Ship shafting reliability evaluation method and system based on multiple working conditions

Abstract

The invention discloses a ship shafting reliability evaluation method and system based on multiple working conditions, and the method comprises the steps: collecting multi-source data of a ship propulsion system through integrating multiple sensors; constructing a dynamic load spectrum based on a time-varying Copula function in combination with parameters such as a stress distribution function and a degradation amount distribution function; a reduced basis finite element method is adopted, experimental strain data are collected, and a reduced-order basis matrix is constructed; based on a three-stage Wiener degradation model, in combination with parameters such as a material degradation rate, activation energy and a temperature average value, the residual life of the ship propulsion system is predicted, and model parameters are adjusted in real time through a Bayesian dynamic updating mechanism; building Hamiltonian through physical constraint reinforcement learning, and optimizing operation parameters of the ship propulsion system in combination with a fault probability threshold value and an energy change rate threshold value; a digital twin engine is utilized to provide real-time visual output. According to the invention, the service life is prolonged to the greatest extent while the system meets the performance requirements; and monitoring and decision making by operators are facilitated.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of ship engineering, and particularly relates to a ship shafting reliability evaluation method and system based on multiple working conditions. BACKGROUND

[0002] With the rapid development of intelligent ship technology, ship propulsion systems are developing towards more efficient, safe and intelligent. However, the traditional sensor configuration scheme based on fault diagnosis and the online monitoring method of system equipment state have been difficult to meet the new requirements of modern intelligent ship operation and maintenance. On the one hand, with the improvement of the complexity of ship system equipment, various state parameter history sensors and state monitoring real-time sensors derived therefrom present the characteristics of "big sensor" such as various forms, large size, multiple dimensions and low value density. It is particularly important to actively perceive and systematically dig the value hidden behind the sensors to realize the evaluation and prediction of the running status of intelligent ships and make treatment strategies for different events. On the other hand, ship propulsion systems are affected by various dynamic loads and environmental factors in actual operation, such as changes in parameters such as speed, torque, vibration, temperature and pressure. These changes directly affect the performance and life of the ship propulsion system. Therefore, a method is needed that can integrate multi-source data to realize real-time monitoring, dynamic evaluation and prediction of the state of the ship propulsion system to improve the safety and economy of the ship.

[0003] At present, some researches at home and abroad have explored the state monitoring and fault diagnosis of ship propulsion systems. For example, a ship residual life prediction method based on multi-source information is proposed. By collecting multiple groups of sensor measurement data, the implicit performance degradation process of infinite divisibility is introduced in the calculation process. Using the state estimation value at the last moment and the linear degradation process of performance, the state estimation value at the next moment closest to the true value can be obtained. According to the state estimation value closest to the true value, the state optimal estimation value at each collection time is obtained by optimizing each past state value. According to the obtained state optimal estimation value, the state value at each future sampling time is estimated, and then compared with the threshold value to predict the residual life of the ship. In addition, the Bayesian network model is used to study the PHM (Prognostics and Health Management) technology of unmanned ship propulsion system. Through the dynamic Bayesian network model, uncertainty information can be effectively processed, and the accuracy of fault prediction can be improved.

[0004] However, the prior art still has certain limitations in processing multi-source data fusion, dynamic load spectrum construction, stress field solving, and residual life prediction. For example, in processing multi-source data, the prior art often lacks deep mining and fusion of data, resulting in insufficient accuracy of the prediction results; in constructing a dynamic load spectrum, the influence of ship body deformation coupling factors and material coupling coefficients is not fully considered, resulting in inaccurate generation of the load spectrum; in solving the stress field, the traditional finite element method has a large amount of calculation and is difficult to meet the real-time requirement; in predicting the residual life, the existing model mostly uses a static model and fails to fully consider the real-time changes of dynamic loads and environmental factors. SUMMARY

[0005] In view of the defects in the prior art, the present application provides a ship shafting reliability evaluation method based on multiple working conditions, comprising the following steps: Step S101, real-time acquisition of ship shafting operation data by a distributed sensor array, wherein the ship shafting operation data is multi-source data; Step S103, fusion of the ship shafting operation data by using a time-varying Copula function to generate a shafting dynamic load spectrum function Ψ(t), wherein the temperature attenuation effect and the pressure-humidity coupling effect are integrated by a nonlinear operator; Step S105, input of Ψ(t) into a shafting-ship coupling finite element model to obtain a maximum stress value σ max ; Step S107, calculation of the shafting residual life L based on an improved three-stage Wiener degradation model, using the following formula to calculate the shafting residual life L: wherein γ is a material degradation rate scalar, ε is a stress sensitivity coefficient scalar, σ max is the maximum stress value, m is a material constant scalar, E a is an activation energy scalar, k B is the Boltzmann constant, T avg is an average temperature scalar, Ψ(t) is a dynamic load spectrum function, erf is an error function, κ is a time attenuation factor scalar, t c is a running-in period threshold, is an indicator function, t is a clock time, and Ti is a cumulative running time; Step S109, real-time updating of the degradation model hyperparameters by a Bayesian Markov chain Monte Carlo algorithm; Step S1011, dynamic adjustment of the shafting speed and load by using a deep reinforcement learning strategy so that the failure probability is lower than a preset threshold; Step S1013, generation of a three-dimensional report integrating a stress cloud map, a life prediction curve, and optimization parameters.

[0006] The ship shafting operation data include rotating speed ω, torque τ, axial vibration acceleration α, radial vibration displacement δ, bearing temperature T, lubricating oil pressure P and environmental humidity H.

[0007] The time-varying Copula function in the step S103 is of Gumbel-Hougaard structure, and a dependent parameter θ(t) is dynamically adjusted according to the hull deformation amount.

[0008] The curvature change rate is measured by the optical fiber strain sensor. θ(t) is calculated. Wherein, k is the material coupling coefficient.

[0009] The shafting-hull coupling finite element model in the step S105 is reduced by using a reduced basis method, and a reduced basis matrix Φ is solved by minimizing Sobolev norms of test strain data and simulation results.

[0010] The improved three-stage Wiener degradation model in the step S107 comprises: An exponential growth drift parameter is used in the running-in period. A stress-driven drift parameter is used in the stable period. A load accumulation drift parameter is used in the failure period. Wherein, τ, κ and ξ are calibrated by bearing material fatigue tests.

[0011] The Bayesian update in the step S109 comprises an adaptive prior distribution, and the precision parameter τ μ The second-order derivative of the reliability index is dynamically calculated. .

[0012] The sensor array in the step S101 comprises a laser displacement meter, a magneto-electric torque meter and a nano pressure-sensitive film, and measurement points are arranged at the thrust bearing, the intermediate bearing and the stern shaft seal according to the ISO 20848 standard.

[0013] The reinforcement learning in the step S1011 adopts a Hamilton constraint mechanism, and a target function satisfies: Wherein Is the shafting energy function.

[0014] The application further provides a ship shafting reliability evaluation system based on multiple working conditions, comprising ​​A multi-source sensing module for collecting ship shafting operation data in real time through a distributed sensor array, the ship shafting operation data being multi-source data; A load spectrum calculation module for fusing the ship shafting operation data using a time-varying Copula function to generate a shafting dynamic load spectrum function Ψ(t), wherein a temperature attenuation effect and a pressure-humidity coupling effect are integrated through a nonlinear operator; A finite element solver for inputting Ψ(t) into a shafting-hull coupled finite element model to obtain a maximum stress value σ max ; A life prediction engine for calculating a shafting residual life L based on an improved three-stage Wiener degradation model, using the following formula to calculate the shafting residual life L: wherein γ is a material degradation rate scalar, ε is a stress sensitivity coefficient scalar, σ max is a maximum stress value, m is a material constant scalar, E a is an activation energy scalar, k B is a Boltzmann constant, T avg is an average temperature scalar, Ψ(t) is a dynamic load spectrum function, erf is an error function, κ is a time attenuation factor scalar, t c is a break-in period threshold, is an indicator function, t is a clock time, and Ti is a cumulative operation time; An updating module for updating degradation model hyperparameters in real time through a Bayesian Markov Chain Monte Carlo algorithm; A reinforcement learning optimizer that dynamically adjusts shafting speed and load using a deep reinforcement learning strategy to make the failure probability lower than a preset threshold; A digital twin visualization terminal for generating a three-dimensional report integrating stress cloud maps, life prediction curves, and optimization parameters.

[0015] Compared with the prior art, the present application has the following advantages: Multi-source data fusion. By integrating various sensors (such as fiber optic encoders, magneto-electric torque meters, MEMS piezoelectric sensors, FBG optical fiber sensors, infrared thermocouple arrays, nano pressure-sensitive film sensors, and capacitive sensors), multi-source data of the ship propulsion system is collected, including speed, torque, vibration signals, temperature, pressure, humidity, and other parameters, achieving comprehensive monitoring of the ship propulsion system status.

[0016] Dynamic load spectrum construction. Based on a time-varying Copula function, combined with stress distribution functions, degradation quantity distribution functions, hull deformation coupling factors, and material coupling coefficients, a dynamic load spectrum is constructed, which can more accurately reflect the load changes of the ship propulsion system under different working conditions.

[0017] Stress field solution. The reduced basis finite element method (RB-FEM) is used to collect experimental strain data, construct a reduced basis matrix, and achieve efficient solution of the stress field of the ship propulsion system. The calculation results meet the CB / Z 208-1983 ship shafting strength calculation specification.

[0018] Remaining life prediction. Based on the three-stage Wiener degradation model, combined with parameters such as material degradation rate, activation energy, and temperature average, the remaining life of the ship propulsion system is predicted, and the model parameters are adjusted in real time through the Bayesian dynamic updating mechanism to improve the accuracy of the prediction.

[0019] Operating parameter optimization. Through physical constraint reinforcement learning, a Hamiltonian is constructed, combined with fault probability threshold and energy change rate threshold, to optimize the operating parameters of the ship propulsion system, ensuring that the system meets the performance requirements while maximizing the service life.

[0020] Visual output. Using a digital twin engine, geometric modeling, data compression, and interactive verification of the ship propulsion system are achieved, providing real-time visual output for operators to monitor and make decisions. BRIEF DESCRIPTION OF DRAWINGS

[0021] The above and other objects, features and advantages of the exemplary embodiments of the present disclosure will be more apparent from the following detailed description taken in conjunction with the accompanying drawings, in which several embodiments of the present disclosure are shown by way of example, and wherein like or corresponding elements show like or corresponding parts, in which: Figure 1 is a flow chart showing a multi-working-condition-based ship shafting reliability evaluation method according to an embodiment of the present application. DETAILED DESCRIPTION

[0022] In order to make the objectives, technical solutions and advantages of the present application clearer, the present application will be described in further detail below with reference to the drawings. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the scope of protection of the present application.

[0023] The terms used in the embodiments of the present application are only for the purpose of describing specific embodiments, and are not intended to limit the present application. The singular forms "a", "an" and "the" used in the embodiments of the present application and the appended claims are also intended to include the plural forms, unless the context clearly indicates otherwise. "Multiple" generally includes at least two.

[0024] It should be understood that, although the terms first, second, third, etc. can be employed in describing the … in the embodiments of the present application, these … should not be limited to these terms. These terms are only used to distinguish one … from another. For example, without departing from the scope of the embodiments of the present application, the first … can also be referred to as the second …, and similarly, the second … can also be referred to as the first ….

[0025] It should be understood that the term "and / or" used herein only describes the association relationship of the associated objects, which means that there can be three relationships, for example, A and / or B can mean that A exists alone, A and B exist together, and B exists alone. In addition, the character " / " herein generally represents that the front and rear associated objects have an "or" relationship.

[0026] Depending on the context, the word "if" as used herein can be interpreted as "when" or "upon" or "in response to a determination" or "in response to a detection". Similarly, depending on the context, the phrase "if determined" or "if detecting (a stated condition or event)" can be interpreted as "when determined" or "in response to a determination" or "when detecting (a stated condition or event)" or "in response to a detection (a stated condition or event)".

[0027] It should also be noted that the terms "comprising", "including", or any other variant thereof are intended to cover non-exclusive inclusion, so that a product or device including a series of elements not only includes those elements, but also includes other elements not explicitly listed, or further includes elements inherent to such product or device. Without more limitations, the element defined by the sentence "including a …" does not exclude the presence of another identical element in the product or device including the element.

[0028] The optional embodiments of the present application will be described in detail below with reference to the accompanying drawings.

[0029] Embodiment one, As Figure 1 shown, the present application discloses a ship shafting reliability evaluation method based on multiple working conditions, comprising the following steps: Step S101, real-time acquisition of ship shafting operation data by distributed sensor array, the ship shafting operation data being multi-source data; Step S103, fusion of the ship shafting operation data by using time-varying Copula function, to generate shafting dynamic load spectrum function Ψ(t), wherein the temperature attenuation effect and the pressure-humidity coupling effect are integrated by a nonlinear operator; Step S105, input of Ψ(t) into the shafting-hull coupling finite element model to obtain the maximum stress value σ max ; Step S107, calculating the shafting residual life L based on the improved three-stage Wiener degradation model, and the shafting residual life L is calculated by the following formula: Wherein, gamma is a material degradation rate scalar, epsilon is a stress sensitivity coefficient scalar, sigma max is a maximum stress value, m is a material constant scalar, E a is an activation energy scalar, k B is a Boltzmann constant, T avg is an average temperature scalar, Psi (t) is a dynamic load spectrum function, erf is an error function, kappa is a time decay factor scalar, t c is a running-in period threshold, is an indicator function, t is a clock time, and Ti is a cumulative running time. Step S109, updating the degradation model hyperparameters in real time through the Bayesian Markov chain Monte Carlo algorithm; Step S1011, dynamically adjusting the shafting speed and load by using a deep reinforcement learning strategy, so that the failure probability is lower than a preset threshold; Step S1013, generating a three-dimensional report of the integrated stress cloud map, the life prediction curve and the optimization parameters.

[0030] Embodiment two, The ship shafting reliability evaluation method based on multiple working conditions provided by the application comprises the following steps: Step S101, collecting ship shafting operation data in real time through a distributed sensor array, wherein the ship shafting operation data is multi-source data; Step S103, fusing the ship shafting operation data by using a time-varying Copula function to generate a shafting dynamic load spectrum function Psi (t), wherein the temperature decay effect and the pressure-humidity coupling effect are integrated through a nonlinear operator; Step S105, inputting Psi (t) into a shafting-hull coupling finite element model to obtain a maximum stress value sigma max ; Step S107, calculating the shafting residual life L based on the improved three-stage Wiener degradation model, and the shafting residual life L is calculated by the following formula: Wherein, gamma is a material degradation rate scalar, epsilon is a stress sensitivity coefficient scalar, sigma max is a maximum stress value, m is a material constant scalar, E a is an activation energy scalar, k B is a Boltzmann constant, T avg is an average temperature scalar, Psi (t) is a dynamic load spectrum function, erf is an error function, kappa is a time decay factor scalar, t c is a running-in period threshold, where t is the clock time, and T is the cumulative running time; Step S109, updating the degradation model hyperparameters in real time through the Bayesian Markov Chain Monte Carlo algorithm; Step S1011, dynamically adjusting the shafting speed and load using a deep reinforcement learning strategy so that the failure probability is below a preset threshold; Step S1013, generating a three-dimensional report of the integrated stress cloud map, life prediction curve, and optimized parameters.

[0031] The ship shafting operation data includes rotational speed ω, torque τ, axial vibration acceleration α, radial vibration displacement δ, bearing temperature T, lubricating oil pressure P, and environmental humidity H.

[0032] The time-varying Copula function of step S103 is of Gumbel-Hougaard structure, and its dependence parameter θ(t) is dynamically adjusted according to the amount of hull deformation.

[0033] The curvature change rate is measured by a fiber optic strain sensor. Calculate θ(t): where k is the material coupling coefficient.

[0034] In this application, the dynamic adjustment of θ(t) is based on the rate of change of the hull deformation. This mechanism embodies the time-varying characteristics of the Copula parameter, i.e., the "time-varying Copula" model. In the time-varying Copula model, the parameter is not fixed, but dynamically adjusted according to the changes in external variables (such as the deformation rate).

[0035] The shafting-hull coupling finite element model in step S105 is reduced by the reduced basis method, and the reduced basis matrix Φ is solved by Sobolev norm minimization of the test strain data and simulation results.

[0036] The coupling finite element model is reduced by the reduced basis method (RBM), which is based on the idea of constructing a low-dimensional subspace to approximate the solution of a high-dimensional problem, thereby significantly reducing the computational complexity. Specifically, the reduced basis matrix is solved by Sobolev norm minimization of the test strain data and simulation results. This process involves the following key steps: Parameterization of the model, representing the partial differential equation (PDE) to be solved in a parameterized form.

[0037] High-fidelity discretization, select a set of parameter values in the parameter space, and perform high-precision finite element solution for each parameter value to obtain a set of high-fidelity solutions. These solutions constitute the "training set" for constructing the reduced basis.

[0038] The basis function construction extracts a set of linearly independent basis functions from the training set by minimizing the Sobolev norm, forming a reduced basis matrix. The choice of Sobolev norm ensures that the basis functions not only approach the function values, but also have good approximation performance on derivatives, thereby improving the overall approximation accuracy.

[0039] In the online phase, for new parameter values, the constructed basis matrix and pre-computed offline information are used to quickly solve low-dimensional problems to obtain approximate solutions. This process is efficient because the cost of solving low-dimensional problems is much lower than that of high-dimensional problems.

[0040] Error estimation To ensure the reliability of the approximate solution, a posteriori error estimator is usually introduced to evaluate the error between the approximate solution and the true solution. This error estimation is based on parameters and has a calculation cost independent of the dimension of the basis function, so it can be efficiently applied to the online phase.

[0041] Coupling strategy In some cases, the model may contain multiple sub-domains or modules, in which case the finite element-simplified approximation basis coupling method can be used. For example, use complete finite elements to describe the region containing local singularities, and use reduced basis methods in smooth regions to ensure the continuity and computational efficiency of the entire model.

[0042] The improved three-stage Wiener degradation model in step S107 includes: The drift parameter increases exponentially during the running-in period ; The drift parameter is stress-driven during the stable period ; The drift parameter is load-accumulative during the failure period , where τ, κ, ξ are calibrated by bearing material fatigue tests.

[0043] The three-stage Wiener degradation model is a mathematical model used to describe the performance changes of products or systems in different degradation stages. Its core idea is to introduce drift parameters (μ) to characterize the degradation rate in different stages. The model divides the degradation process into three stages: running-in period, stable period and failure period. The drift parameters in each stage have different mathematical expressions, and the values of these parameters depend on material properties, stress levels and load accumulation.

[0044] The Bayesian update in step S109 includes an adaptive prior distribution with precision parameter τ μ The second derivative of the reliability index is dynamically calculated as: , where is a precision parameter, used to control the width of the prior distribution; is a reliability indicator, representing the reliability of the system at time t; t i is the i-th time point; n is the total number of time points. This formula shows that the size of the precision parameter depends on the curvature of the reliability indicator (i.e., the absolute value of the second derivative), thus achieving adaptive updating.

[0045] Bayesian updating is a method based on Bayes' theorem, which updates the posterior distribution of model parameters by combining prior information and observed data. In the adaptive Bayesian framework, the prior distribution is not fixed, but dynamically adjusted according to new data or environmental changes. This adaptive mechanism can improve the flexibility and robustness of the model, especially for nonlinear or high-uncertainty problems.

[0046] In this step S5, the dynamic calculation of the precision parameter reflects this adaptive characteristic. By calculating the second derivative of the reliability indicator, the reliability change trend of the system at different time points can be captured. For example, if the reliability indicator changes dramatically (i.e., the second derivative is large) in a certain time period, the precision parameter will also increase accordingly, making the prior distribution more concentrated and reducing the dependence on new data; conversely, if the reliability indicator changes smoothly, the precision parameter will decrease, making the prior distribution more extensive and increasing the adaptability to new data.

[0047] The core of the Bayesian method is to update the posterior distribution by combining the prior distribution and the likelihood function. In the adaptive Bayesian framework, the update of the prior distribution not only depends on historical data, but also depends on the current environmental state or system behavior. For example, in clinical trials, Bayesian methods can dynamically adjust sample size, grouping ratio, or trial design according to interim trial data, thereby improving trial efficiency and ethics. Similarly, in reliability analysis, by dynamically updating the precision parameter , the reliability change of the system can be better reflected, thereby optimizing the prediction performance of the model.

[0048] The sensor array in step S101 includes a laser displacement meter, a magneto-electric torque meter, and a nano pressure-sensitive film. Measurement points are arranged at the thrust bearing, intermediate bearing, and stern shaft seal according to the ISO 20848 standard.

[0049] A laser displacement meter is a high-precision non-contact measuring device based on the principle of laser interference, capable of accurately measuring the displacement, vibration, thickness and other parameters of an object. It consists of a laser, a light emitting lens, a light receiver, a linear image sensor and a measurement circuit. When the laser beam shines on the surface of the measured object, the reflected light enters the linear image sensor through the light receiver lens, thereby detecting the position change of the object. Laser displacement meters have high resolution, high frequency response and good stability, and are suitable for precision measurement in industrial and scientific fields. For example, the LD-40 laser displacement sensor can measure static water level, structural deformation, settlement, etc., and supports RS485 digital output, making it easy to integrate into an automation system. In the ship power system, laser displacement meters can be used to measure the axial displacement and vibration of the thrust bearing, intermediate bearing and stern shaft seal to evaluate their running state.

[0050] A magneto-elastic torque meter is a precision instrument used to measure the output torque of rotating machinery, widely used in electric motors, engines, internal combustion engines, fans, water pumps, gearboxes and other equipment. Its working principle is based on electromagnetic induction, which calculates the torque value by measuring the electromagnetic signal generated by the relative motion between the rotor and the stator. Magneto-elastic torque meters have high precision, wide frequency band and strong anti-interference ability, suitable for vibration measurement tasks in strong magnetic fields. In the ship power system, magneto-elastic torque meters can be used to measure the torque of the main engine output shaft to evaluate its power transmission efficiency and running stability.

[0051] A nano pressure-sensitive film is a sensor based on piezoelectric materials such as PVDF / MWNT nanocomposites, which can convert mechanical stress into electrical signals to measure physical quantities such as vibration and pressure. It has high sensitivity, wide frequency band and good environmental adaptability, suitable for high-frequency vibration detection and micro-deformation measurement. In the ship power system, nano pressure-sensitive films can be used to measure the local stress and vibration of the thrust bearing, intermediate bearing and stern shaft seal to evaluate their fatigue life and running state.

[0052] ISO 20848 is an international standard developed by the International Organization for Standardization (ISO) for measuring the vibration and noise of ship propulsion systems, aiming to standardize the performance evaluation and fault diagnosis of ship power systems. The standard specifies the selection of measurement points, sensor arrangement, data acquisition and analysis methods to ensure the comparability and reliability of measurement results. In practical applications, arranging measurement points according to the ISO 20848 standard can ensure that the measurement data of laser displacement meters, magneto-elastic torque meters and nano pressure-sensitive films meet uniform technical requirements, thereby improving the overall performance of the monitoring system.

[0053] According to the ISO 20848 standard, the measuring points should be arranged at the thrust bearing, intermediate bearing and stern shaft seal, which are key support and transmission components in the ship power system, and their operating conditions directly affect the performance and safety of the whole machine. Laser displacement meter is used to measure axial displacement and vibration, magneto-electric torque meter is used to measure torque output, and nano pressure sensitive film is used to measure local stress and vibration. Through the cooperative measurement of multiple sensors, the running state of the ship power system can be comprehensively evaluated, potential faults can be found in time and preventive maintenance can be carried out.

[0054] wherein the reinforcement learning of step S1011 adopts a Hamilton constraint mechanism, and a target function satisfies: wherein is an energy function of the shafting.

[0055] A Hamilton system is a kind of dynamic system with energy conservation characteristics, and its evolution follows Hamilton equation. In reinforcement learning, the introduction of Hamilton constraint mechanism can integrate the physical characteristics (such as energy conservation) of the system into the learning process. In reinforcement learning, policy gradient methods (such as PPO, SAC, etc.) are usually used to optimize the policy to maximize the expected return. The introduction of Hamilton constraint mechanism can directly integrate these physical constraints into the objective function of reinforcement learning, so as to ensure that the learned policy not only maximizes the reward, but also meets the characteristics of the physical system.

[0056] wherein the Hamilton constraint mechanism comprises a thermal deformation compensation module, which automatically triggers lubrication flow adjustment when the bearing temperature gradient exceeds a threshold value: wherein η is an adjustment coefficient, T th is a temperature gradient threshold.

[0057] Hamilton mechanics is a mathematical framework for describing the dynamic behavior of physical systems, widely used in classical mechanics and modern control theory. In a constrained Hamilton system, the behavior of the system is limited by constraint conditions, which are usually related to the conservation of energy, momentum, etc. of the system. In this invention, a Hamilton constraint mechanism is introduced into the thermal deformation compensation system, which realizes the dynamic adjustment of the lubrication system by constructing a feedback control model based on the temperature gradient.

[0058] Specifically, this mechanism monitors the bearing temperature gradient When it exceeds the set threshold T th , the system will automatically adjust the lubrication flow Q oilA Sigmoid function in the adjustment formula is used to achieve nonlinear adjustment, so that the lubrication flow increases rapidly when the temperature gradient approaches the threshold value, and maintains a small flow when the temperature gradient is below the threshold value, achieving smooth transition and efficient control.

[0059] In practical applications, the thermal deformation compensation module is usually implemented in the following ways: Sensor and feedback control, by arranging temperature sensors, real-time monitoring of temperature distribution in key parts, and calculating temperature gradient When the temperature gradient exceeds the threshold value, the control system will adjust the lubrication flow according to the preset adjustment strategy (such as Sigmoid function) to compensate for thermal deformation.

[0060] Custom compensation algorithm, in the numerical control system, users can embed custom thermal deformation compensation modules, develop multi-temperature sensor multi-element compensation algorithms, and achieve high-precision compensation functions for specific machine tools.

[0061] Structural design compensation, in some cases, thermal deformation can be absorbed or released through structural design, such as using corrugated pipes or spring structures, or using different thermal expansion coefficient material combinations to reduce the impact of thermal deformation on the system.

[0062] Online detection and feedback, real-time monitoring of workpiece thermal deformation error through online detection system, and compensation through feedback control to improve machining accuracy.

[0063] Among them, the visual report integrates a digital twin engine, which realizes: Real-time rendering of stress contours; Dynamic update of life prediction curves; Optimization parameter compliance automatic verification.

[0064] The digital twin engine combined with WebGL technology can render the stress distribution of the physical system in real time. WebGL is a Web-based 3D graphics rendering technology that allows direct rendering of high-quality 3D models in browsers without additional plugins. By inputting sensor data, simulation data, etc. into the digital twin model, the system can generate real-time stress contours to help engineers and maintenance personnel intuitively understand the stress state of equipment or structures during operation. This real-time rendering capability is due to the efficient rendering mechanism of WebGL and GPU acceleration technology, making it possible to smoothly display complex 3D scenes even on ordinary devices.

[0065] Digital twin technology can dynamically predict the service life of equipment by integrating historical operation data, sensor data, and simulation models. The system will continuously update the life prediction curve based on the operating parameters of the equipment (such as temperature, pressure, vibration, etc.) and material aging characteristics. This dynamic updating mechanism relies on the real-time data interaction capability of the digital twin model and the prediction algorithm based on machine learning. In this way, enterprises can early warning potential equipment failures, optimize maintenance plans, and extend equipment service life.

[0066] The digital twin system also has an automatic verification function, which can check the compliance of the optimized parameters of the equipment. In the digital twin environment, the system can automatically retrieve relevant standard parameters and compare them with the actual operating parameters of the equipment to determine whether they meet the specification requirements. This automatic verification function not only improves work efficiency but also reduces the compliance risks caused by human errors.

[0067] Example Three, In one embodiment, in step S101, multi-source data is collected.

[0068] The sensor configuration is as follows: Rotational speed ω, fiber optic encoder (accuracy ±0.1 rad / s), installed at the output flange of the main machine; Torque τ, magnetic torque meter (range ), integrated strain gauge bridge; The vibration signal includes: Axial acceleration α, MEMS piezoelectric sensor (frequency response 0.5-10 kHz); Radial displacement δ, FBG optical fiber sensor (resolution ±0.01 mm); Bearing temperature T, infrared thermocouple array (spatial resolution 2 mm); Lubricating oil pressure P, nano pressure-sensitive film sensor (range 0-50 MPa); Environmental humidity H, capacitive sensor (±1% RH).

[0069] In step S103, a dynamic load spectrum is constructed.

[0070] (1) The time-varying Copula function is implemented using the following formula: where represents the stress distribution function; represents the degradation quantity distribution function, which is the cumulative fatigue quantity of the shaft material; represents the hull deformation coupling factor, k represents the material coupling coefficient, for example, the steel hull takes 0.25, represents the curvature change rate, which is measured in real time by the fiber strain sensor.

[0071] (2) Load spectrum is generated by the following formula: Where the key coefficients are as follows:

[0072] Table I i represents the sensor number, t i represents the discrete time sampling point.

[0073] Step S103, stress field solving.

[0074] The reduced basis finite element method (RB-FEM) is implemented by the following formula: , I m is the experimental grid interpolation operator.

[0075] Where the complete expression of the reduced stiffness matrix is as follows: .

[0076] The parameters are explained as follows:

[0077] Table II Where the reduced basis is constructed by the following steps: Reference strain gage measurement method is used to collect experimental strain data ; The error is minimized by Sobolev norm to generate the basis matrix .

[0078] The stress output is as follows: .

[0079] Step S107, residual life prediction.

[0080] The three-stage Wiener degradation model is calculated by the following formula: , represents the material inherent degradation rate, which is calibrated by fatigue experiment; m represents the stress sensitivity index, which is obtained by fitting the acceleration degradation experiment; represents the low-frequency load filtering coefficient, which is obtained by vibration spectrum analysis; T avg represents the average temperature of the bearing, which is monitored in real time by infrared array.

[0081] Stage control logic:

[0082] Table III Parameter calibration: (material degradation rate, fatigue test acquisition); (activation energy, diffusion energy barrier of microscopic defects in bearing steel); : error function filter low frequency load fluctuations .

[0083] Step S109, Bayesian dynamic update.

[0084] Adaptive prior distribution design as follows: .

[0085] Update mechanism: When the second derivative of the reliability index mutates (when ), the prior distribution variance is automatically reduced.

[0086] Through MCMC sampling times, output the posterior distribution .

[0087] Step S1011, operating parameter optimization.

[0088] Physical constraint reinforcement learning, using the following formula: , where, represents the failure probability of the propulsion system; represents the Hamilton energy change rate threshold, which is set based on the material fatigue limit.

[0089] The Hamiltonian is constructed using the following formula: .

[0090] Control strategy as follows: Viscosity adjustment, based on temperature gradient threshold T th Dynamic control of flow. When threshold (threshold), trigger lubrication flow adjustment: . Step S1013, visual output.

[0091] The digital twin engine is implemented using the following steps:

[0092] Geometric modeling, which parameterizes the NURBS surface of the hull (refers to the three-dimensional design standard); ​Data compression, which uses proper orthogonal decomposition (POD) to compress the stress field to 5% of the original size; Interactive verification, which uses the WebGL engine for real-time rendering and embeds ISO 20848 compliance verification algorithms.

[0093] Example four, The application also provides a ship shafting reliability evaluation system based on multiple working conditions, comprising A multi-source sensing module is configured to collect ship shafting operation data in real time through a distributed sensor array, wherein the ship shafting operation data is multi-source data. A load spectrum calculation module is configured to fuse the ship shafting operation data by using a time-varying Copula function to generate a shafting dynamic load spectrum function Ψ(t), wherein a temperature attenuation effect and a pressure-humidity coupling effect are integrated through a nonlinear operator. A finite element solver is configured to input Ψ(t) into a shafting-hull coupled finite element model to obtain a maximum stress value σ max ; A life prediction engine is configured to calculate the shafting residual life L based on an improved three-stage Wiener degradation model, and the shafting residual life L is calculated by using the following formula: wherein γ is a material degradation rate scalar, ε is a stress sensitivity coefficient scalar, σ max is a maximum stress value, m is a material constant scalar, E a is an activation energy scalar, k B is a Boltzmann constant, T avg is an average temperature scalar, Ψ(t) is a dynamic load spectrum function, erf is an error function, κ is a time attenuation factor scalar, t c is a running-in period threshold, is an indicator function, t is a clock time, and Ti is a cumulative running time. An updating module is configured to update the degradation model hyperparameters in real time by using a Bayesian Markov chain Monte Carlo algorithm. A reinforcement learning optimizer is configured to dynamically adjust the shafting speed and load by using a deep reinforcement learning strategy so that the failure probability is lower than a preset threshold. A digital twin visualization terminal is configured to generate a three-dimensional report integrating a stress cloud map, a life prediction curve, and optimized parameters.

[0094] The multi-source sensing module is responsible for collecting data from multiple sensors, which may come from different physical devices or environments. After preprocessing, the data is sent to the data association module for integration and analysis. The design of the multi-source sensing module aims to improve the accuracy and real-time performance of target tracking.

[0095] The load spectrum calculation module is used to calculate and manage the load spectrum, which supports rainflow counting of time-domain load spectrum obtained through tests and provides customization and correction functions. This module also supports automatic output of maintenance calculation results to ensure the accuracy of the calculation.

[0096] The finite element solver is used to simulate the behavior of a structure under stress by breaking it down into small elements (finite elements) and calculating stress, strain, and other parameters for each element.

[0097] The life prediction engine is used to evaluate the service life of a product or system, usually based on S-N curves (fatigue life curves) and load spectrum data. This module predicts the service life of a structure under specific working conditions by analyzing its fatigue characteristics.

[0098] The reinforcement learning optimizer is an optimization tool based on reinforcement learning algorithms, used to train an efficient optimizer. It interacts with the environment, constantly adjusting the strategy to maximize rewards, to achieve the optimal solution.

[0099] The digital twin visualization terminal is responsible for displaying the running state, performance, and health indicators of the digital twin system to users in a graphical manner. Through real-time cloud rendering technology, users can access real three-dimensional content and perform interactive operations on lightweight terminals, achieving the goal of terminal lightweight and unlimited cloud.

[0100] Each module is interconnected through a time-sensitive network (TSN), which is an industrial Ethernet standard that supports deterministic data transmission, ensuring efficient and low-latency data transmission in industrial IoT. Through TSN, communication between modules is more stable and reliable, avoiding data conflicts and latency issues in traditional networks.

[0101] These modules together form a highly integrated intelligent system that can achieve full-process automation from data collection, processing, analysis to optimization and visualization. Through the interconnection of the time-sensitive network, efficient data exchange and collaborative work between modules can be achieved, thereby improving the overall performance and reliability of the system.

[0102] Example five, The present disclosure provides a non-volatile computer storage medium, which stores computer executable instructions. The computer executable instructions can execute the method steps as described in the above embodiments.

[0103] Note that the computer readable medium described above can be a computer readable signal medium or a computer readable storage medium or any combination thereof. The computer readable storage medium can be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus or device, or any suitable combination of the foregoing. More specific examples of the computer readable storage medium can include, but are not limited to, an electrical connection having one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the disclosure, the computer readable storage medium can be any tangible medium that contains or stores a program used by or in connection with an instruction execution system, apparatus or device. In the disclosure, the computer readable signal medium can include a computer readable program code propagated on or through a computer readable medium, in baseband or as part of a carrier wave. The computer readable signal medium can take a variety of forms, including but not limited to, electro-magnetic, optical, or any suitable combination of the foregoing. The computer readable signal medium can be any computer readable medium that can be used to carry or store computer readable program code for use by or in connection with an instruction execution system, apparatus or device. The program code contained on the computer readable medium can be transmitted or received over any suitable medium, including but not limited to, wire, cable, fiber optic, RF (radio frequency), or any suitable combination of the foregoing.

[0104] The computer readable medium described above can be included in the electronic device described above; alternatively, the computer readable medium can exist as a separate entity in which the electronic device is incorporated.

[0105] The computer program code for carrying out operations of the disclosure can be written in one or more programming languages or combinations of languages including object oriented, such as Java, Smalltalk, C++, and conventional procedural, such as the "C" programming language or similar programming languages. The program code can execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the latter scenario, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or the connection can be made to an external computer (for example, through the Internet using an Internet Service Provider).

[0106] The computer program product of the present disclosure can be a computer program embodied on a non-transitory computer readable medium. When the program runs on a computer, the flowchart and / or block diagram in the flowchart and / or block diagram can be implemented.

[0107] The units described in the embodiments of the present disclosure can be implemented by software or by hardware. In some cases, the names of the units do not constitute a limitation on the units themselves.

[0108] The above describes the preferred embodiments of the present disclosure, which aims to make the spirit of the present disclosure clearer and easier to understand, and is not intended to limit the present disclosure. Any modifications, replacements, improvements made within the spirit and principle of the present disclosure shall be included in the protection scope of the appended claims of the present disclosure.

Claims

1. A method for assessing the reliability of ship shafting based on multiple operating conditions, characterized in that, Includes the following steps: Step S101: Collect ship shafting operation data in real time through a distributed sensor array. The ship shafting operation data is multi-source data. Step S103: The ship shafting operation data is fused using a time-varying Copula function to generate a shafting dynamic load spectrum function Ψ(t), wherein the temperature decay effect and the pressure-humidity coupling effect are integrated through a nonlinear operator; Step S105: Input Ψ(t) into the shaft-hull coupled finite element model to obtain the maximum stress value σ. max ; Step S107: Calculate the remaining life L of the shaft system based on the improved three-stage Wiener degradation model. The remaining life L of the shaft system is calculated using the following formula: Where γ is the scalar of material degradation rate, ε is the scalar of stress sensitivity coefficient, and σ max The maximum stress value is given by E, where m is a material constant scalar, and E is the stress value. a To activate the scalar energy, k B T is the Boltzmann constant. avg Let Ψ(t) be the average temperature scalar, Ψ(t) be the dynamic load spectrum function, erf be the error function, κ be the time decay factor scalar, and t be... c This is the break-in period threshold. Let t be the indicator function, t be the clock time, and Ti be the cumulative running time; Step S109: Update the hyperparameters of the degradation model in real time using the Bayesian Markov chain Monte Carlo algorithm; Step S1011: Dynamically adjust shaft speed and load using deep reinforcement learning strategy to reduce failure probability to a preset threshold. Step S1013: Generate a three-dimensional report of integrated stress cloud map, life prediction curve and optimization parameters.

2. The method as described in claim 1, characterized in that, The ship shafting operating data includes rotational speed ω, torque τ, axial vibration acceleration α, radial vibration displacement δ, bearing temperature T, lubricating oil pressure P, and ambient humidity H.

3. The method as described in claim 1, characterized in that, The time-varying Copula function in step S103 is a Gumbel-Hougaard structure, and its dependent parameter θ(t) is dynamically adjusted with the amount of hull deformation.

4. The method as described in claim 3, characterized in that, in, Specifically, the rate of curvature change is measured using a fiber optic strain sensor. Calculate θ(t): , where k is the material coupling coefficient.

5. The method as described in claim 1, characterized in that, In step S105, the shaft-hull coupled finite element model is reduced in order using the reduced basis method. The reduced basis matrix Φ is solved by minimizing the Sobolev norm of the experimental strain data and simulation results.

6. The method as described in claim 1, characterized in that, The improved three-stage Wiener degradation model in step S107 includes: Exponential growth drift parameter was used during the break-in period. ; Stress-driven drift parameters are used during the stabilization period. ; Failure time uses load cumulative drift parameter τ, κ, and ξ are calibrated through bearing material fatigue tests.

7. The method as described in claim 1, characterized in that, The Bayesian update in step S109 includes an adaptive prior distribution with an accuracy parameter τ. μ Reliability indicators Dynamic calculation of the second derivative: 。 8. The method as described in claim 1, characterized in that, The sensor array in step S101 includes a laser displacement meter, a magnetoelectric torque meter, and a nano-pressure-sensitive film, with measurement points arranged at the thrust bearing, intermediate bearing, and stern shaft seal according to ISO 20848 standard.

9. The method as described in claim 1, characterized in that, The reinforcement learning in step S1011 employs a Hamiltonian constraint mechanism, and the objective function satisfies: ,in This is the axis energy function.

10. A multi-condition-based ship shafting reliability assessment system, including... A multi-source sensing module is used to collect real-time ship shafting operation data through a distributed sensor array, wherein the ship shafting operation data is multi-source data. The load spectrum calculation module is used to fuse the ship shafting operation data using a time-varying Copula function to generate a dynamic load spectrum function Ψ(t) for the shafting, where the temperature decay effect and the pressure-humidity coupling effect are integrated through a nonlinear operator. A finite element solver is used to input Ψ(t) into a shaft-hull coupled finite element model to obtain the maximum stress value σ. max ; The life prediction engine calculates the remaining life L of the shaft system based on an improved three-stage Wiener degradation model, using the following formula: Where γ is the scalar of material degradation rate, ε is the scalar of stress sensitivity coefficient, and σ max The maximum stress value is given by E, where m is a material constant scalar, and E is the stress value. a To activate the scalar energy, k B T is the Boltzmann constant. avg Let Ψ(t) be the average temperature scalar, Ψ(t) be the dynamic load spectrum function, erf be the error function, κ be the time decay factor scalar, and t be... c This is the break-in period threshold. Let t be the indicator function, t be the clock time, and Ti be the cumulative running time; The update module is used to update the hyperparameters of the degenerate model in real time using the Bayesian Markov chain Monte Carlo algorithm; The reinforcement learning optimizer uses deep reinforcement learning strategies to dynamically adjust shaft speed and load, so that the failure probability is lower than a preset threshold. A digital twin visualization terminal that generates 3D reports integrating stress cloud maps, life prediction curves, and optimization parameters.

Images