A real-time prediction and reliability evaluation method for lock bottom hinge mushroom head wear

Abstract

The present application belongs to the technical field of waterway engineering, and discloses a ship lock bottom pivot wear real-time prediction and reliability evaluation method, comprising: establishing a finite element simulation model of the ship lock miter gate leaf and bottom pivot, and obtaining bottom pivot wear law data; based on the finite element simulation data, a Bayesian neural network proxy model is constructed to realize efficient prediction of the bottom pivot wear amount; combined with the proxy model prediction result, a bottom pivot performance degradation evaluation model based on the Gamma process is established, and the sequence Bayesian method is used to real-time correct the model parameters, and dynamically update the bottom pivot wear reliability and residual life. The present application can effectively solve the problems of underwater concealment of the ship lock bottom pivot, difficulty in online monitoring and deficiency of the traditional reliability evaluation method, improve the bottom pivot wear prediction performance and the reliability of the reliability evaluation, and provide intelligent protection for safe and efficient operation of the ship lock.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of water conservancy projects, and particularly relates to a ship lock gate bottom pivot operation maintenance and safety evaluation technology, and specifically discloses a ship lock bottom pivot mushroom head wear real-time prediction and reliability evaluation method. BACKGROUND

[0002] As a key facility of water transport engineering, the safe and reliable operation of the ship lock plays a crucial role in ensuring smooth navigation. However, during operation, various types of failures often occur due to factors such as wear, fatigue, and corrosion of structural components, leading to long-term navigation disruption and significant economic losses. Therefore, in-depth research on the reliability of ship lock structures is of great practical significance in improving operational safety and reducing maintenance costs. Among the many ship lock structures, the bottom pivot, as the core operating component and important support of the miter gate, is subjected to intermittent low-speed heavy loading and mud pollution in a complex underwater hidden environment. Its working conditions are special, and it is prone to wear between friction pairs, making it difficult to form a stable fluid lubrication oil film. Severe wear can cause the gate shaft column to tilt, even leading to gate instability, posing a serious threat to operational safety. In addition, the special underwater hidden environment of the bottom pivot makes it difficult to monitor and evaluate its wear state in real time, making post-failure maintenance work extremely difficult and costly. Although the existing ship lock gate design specifications mostly use the traditional allowable stress method, this method does not fully consider the influence of component type, design parameters, and time variability during service on structural reliability.

[0003] Currently, research on the mushroom head wear of the bottom pivot mainly focuses on the design level, such as structure type optimization, lubrication method improvement, and friction pair material matching. However, there is relatively little research on the real-time prediction and safety reliability evaluation of the mushroom head wear state of the ship lock bottom pivot that has been put into operation. In the field of gate reliability research, although there are achievements on single components such as the main beam, there is still a lack of research on the reliability of the mushroom head wear failure of the ship lock bottom pivot. In addition, traditional reliability evaluation methods have shortcomings in utilizing historical degradation data, and face the problems of large repeated calculation amount and low calculation efficiency during dynamic updating, making it difficult to meet real-time requirements. In summary, the current ship lock bottom pivot mushroom head wear problem lacks effective online monitoring and prediction means, the reliability evaluation method is not perfect and the calculation efficiency needs to be improved, and a new method integrating multi-disciplinary theory and data-driven technology is urgently needed to break through the bottleneck of the reliability research of the ship lock miter gate bottom pivot mushroom head wear.

[0004] CN108318363A discloses a detection device for bottom pivot mushroom head wear and a detection method thereof. The device contacts the mushroom head surface through a mechanical measuring needle, obtains surface data and reconstructs a spherical surface, and then compares the difference between the spherical surfaces before and after wear to evaluate the wear characteristics. This method can detect wear without disassembling the equipment. However, this method is essentially an offline detection technology that relies on manual operation and measurement on site, and cannot achieve continuous and real-time monitoring of the bottom pivot wear state, making it difficult to capture dynamic changes in the wear process. In addition, the method obtains data through contact measurement, and the operation process is relatively complicated, and the measurement accuracy is affected by the underwater environment (such as silt and attachments).

[0005] CN111859732A discloses a ship lock gate and a method for automatically monitoring the damage degree of a supporting running part thereof. The method indirectly monitors underwater hidden parts including the bottom pivot by arranging sensors and obtaining vibration response parameters in real time, and using a vibration parameter response relationship model. This method achieves automatic monitoring of underwater hidden parts, but it is difficult to directly and quantitatively determine the wear amount of the bottom pivot mushroom head, and the evaluation results have uncertainties. In addition, the mapping relationship between vibration response and wear state is complex, and a precise model needs to be established, which requires high generalization ability of the model. Furthermore, the above methods do not combine dynamic monitoring of the wear state with long-term reliability and residual life prediction of the structure, and cannot provide prospective guidance for maintenance decisions of the ship lock. Therefore, the existing technology has obvious deficiencies in real-time and quantitative prediction of bottom pivot wear and dynamic integration with reliability evaluation.

[0006] The existing technology has problems in offline measurement, inability to monitor in real time, inability to directly and quantitatively determine the wear amount, and lack of dynamic integration with reliability evaluation in monitoring the wear of the bottom pivot mushroom head of the ship lock. The present application is proposed to solve the problems of difficulty in real-time prediction of the wear state of the bottom pivot mushroom head of the existing ship lock, imperfect reliability evaluation method, and low calculation efficiency. SUMMARY

[0007] This section is intended to summarize some aspects of the embodiments of the present application and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section and the abstract and title of the specification to avoid obscuring the purpose of this section, the abstract and the title. Such simplifications or omissions cannot be used to limit the scope of the present application.

[0008] In view of the above-mentioned existing problems, the present application is proposed.

[0009] To solve the above technical problems, the present application provides the following technical solutions: a finite element simulation model of the friction pair of the miter gate leaf and the bottom pivot of the ship lock is established, the dynamic simulation of the friction contact is considered, the dynamic wear of the mushroom head of the bottom pivot in the opening and closing process of the miter gate is simulated, so as to obtain the wear amount of the mushroom head of the bottom pivot under different working conditions and the corresponding structural dynamic response simulation data;

[0010] Based on the obtained finite element simulation data, a Bayesian neural network proxy model, i.e. a mushroom head wear state proxy model, is constructed and trained based on the Bayesian neural network prediction algorithm, and the mushroom head wear state proxy model is used to efficiently predict the wear amount of the mushroom head of the bottom pivot according to the structural dynamic response simulation data;

[0011] Based on the Gamma random process theory, a mushroom head performance degradation evaluation model is established to describe the long-term random process of wear evolution over time, and the model parameters are initialized and estimated using historical data;

[0012] In combination with the real-time monitored structural dynamic response data, the current wear amount is output by the mushroom head wear state proxy model, and then the wear amount is used as new observation data, and the sequence Bayesian method is used to real-time correct the parameters of the mushroom head performance degradation evaluation model, and the reliability of the mushroom head wear failure and the remaining life prediction result are dynamically updated.

[0013] As a preferred scheme of the ship lock bottom pivot mushroom head wear real-time prediction and reliability evaluation method, the finite element simulation model uses the Lagrange contact algorithm to realize the friction simulation of the mushroom head, and the Archard wear model is used to calculate the wear amount of the mushroom head of the bottom pivot, and the Archard wear model is expressed as:

[0014]

[0015] Wherein, is the wear volume, is the dimensionless wear coefficient, is the normal load, is the relative sliding distance, is the material hardness.

[0016] As a preferred scheme of the ship lock bottom pivot mushroom head wear real-time prediction and reliability evaluation method, the input features of the Bayesian neural network proxy model are selected from one or more physical quantity feature combinations of the stress, strain, vibration response, acoustic emission signal, driving mechanism motor current, gate operation water level and the like of the gate leaf and the supporting rotating part, and the output feature is the cumulative wear amount of the mushroom head.

[0017] As a preferred scheme of the ship lock bottom hinge mushroom head wear real-time prediction and reliability evaluation method, the Bayesian neural network proxy model adopts a deterministic approximation method of a variational inference algorithm for training and prediction, converts a solution of a hidden parameter posterior problem into a variational optimization problem of maximizing an evidence lower bound ELBO, finds an optimal posterior distribution solution through iteration, realizes rapid prediction of the bottom hinge wear state, and the formula of the ELBO is:

[0018]

[0019] Among them, ELBO is an objective function that needs to be maximized in variational inference, is the variational approximation distribution solves the expectation, is the finite element simulation data of the input characteristics of the proxy model, Z is a hidden variable, that is, the weight and bias model parameters of the Bayesian neural network, is a variational approximation distribution for approximating the true posterior distribution , is a likelihood function of the finite element simulation data when the hidden variable Z is given, is a prior distribution of the hidden variable Z, is the Kullback-Leibler divergence between the variational approximation distribution and the prior distribution of the hidden variable .

[0020] As a preferred scheme of the ship lock bottom hinge mushroom head wear real-time prediction and reliability evaluation method, the probability density function of the cumulative wear amount of the mushroom head performance degradation evaluation model is:

[0021]

[0022] Among them, is the probability density of the cumulative wear amount at time , is the cumulative wear amount at time , is the shape parameter of the Gamma distribution, which is a function of time, indicating the degree of cumulative wear, is the scale parameter of the Gamma distribution, reflecting the average size of the wear amount, e is the base of the natural logarithm, is the Gamma function.

[0023] As a preferred scheme of the ship lock bottom hinge mushroom head wear real-time prediction and reliability evaluation method, the initial parameters of the mushroom head performance degradation evaluation model The estimation is performed using the maximum likelihood estimation method, the objective function of which is:

[0024]

[0025] in, Let be the likelihood function. Represents the parameters of the Gamma stochastic process model, i.e. and , Historical observation data used for parameter estimation In time Observed cumulative wear For given parameters Cumulative wear during the Gamma stochastic process The probability density function.

[0026] As a preferred embodiment of the real-time prediction and reliability assessment method for the wear of the mushroom head of the lock bottom pivot as described in this invention, the sequential Bayesian method modifies the parameters of the performance degradation assessment model in real time based on Bayes' theorem. Iterative updates are performed, and their mathematical expression is as follows:

[0027]

[0028] in, In order to observe real-time monitoring data After that, parameters The posterior probability distribution, For given parameters In the case of the latest monitoring data The likelihood function, Let be the posterior probability distribution of the previous time step, and be the current updated prior probability distribution. Indicates proportionality. These are the parameters for the Gamma stochastic process model.

[0029] As a preferred embodiment of the real-time prediction and reliability assessment method for the wear of the mushroom head of the lock bottom pivot as described in this invention, after dynamically updating the reliability and remaining life prediction results of the mushroom head wear failure, the method further includes formulating and implementing a set of predictive maintenance strategies for the mushroom head, specifically including the following steps:

[0030] Calculate the cumulative distribution function of remaining lifetime: Based on the real-time updated mushroom head performance degradation assessment model, a predefined wear failure threshold is set. Calculate at the current time After that, remaining lifespan cumulative distribution function , which represents the probability that the cumulative wear of the bottom hinge will first reach or exceed the failure threshold at future time ;

[0031] Establish a predictive maintenance strategy objective function: establish a function that aims to minimize the long-term expected total cost , which is used to evaluate the economic efficiency of performing preventive maintenance at future maintenance opportunities, in the form of:

[0032]

[0033] where, is the cost of performing one preventive maintenance, is the cost of repairing after failure (usually ), is the calculated probability of failure before time , and is the probability of not failing before time ;

[0034] Determine the optimal preventive maintenance opportunity: determine the optimal preventive maintenance opportunity by solving the optimization problem , that is, find the maintenance time point that makes the objective function defined in the previous step achieve the minimum value :

[0035]

[0036] The optimal preventive maintenance opportunity provides the best maintenance decision window based on cost-benefit analysis.

[0037] As a preferred solution of the ship lock bottom hinge mushroom head wear real-time prediction and reliability evaluation method described in the present application, the step further comprises formulating and executing a maintenance strategy, wherein:

[0038] According to the calculated optimal preventive maintenance opportunity , combined with the current running time and state of the bottom hinge, a maintenance suggestion is generated;

[0039] When the actual running time approaches , the system triggers a maintenance alarm to guide relevant personnel to perform preventive maintenance, and after the maintenance is completed, the maintenance data is fed back to the historical database for the initialization and update of the next round of model, forming a continuous optimization closed-loop management process.

[0040] ​​The beneficial effects of the present application: the present application establishes a reliable wear-response prior correlation through finite element friction contact dynamics, constructs a Bayesian neural network proxy model to realize the probabilistic rapid estimation of wear amount, obtains a reliability / RUL analytical framework directly related to the failure threshold value by means of Gamma random process, and realizes continuous adaptive correction facing field data by means of sequential Bayesian, thereby forming a real-time, probabilistic, interpretable, deployable reliability evaluation and life prediction system facing the bottom pivot of the miter gate; realize online visualization and end-to-end quantitative evaluation of wear state, incorporate uncertainty into decision-making, support maintenance resource optimization and accurate planning of shutdown window, and significantly reduce dependence on offline disassembly and large-scale simulation. BRIEF DESCRIPTION OF DRAWINGS

[0041] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced as follows. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor. Among them:

[0042] Figure 1 The flowchart of the miter gate mushroom head wear real-time prediction and reliability evaluation method shown in the present application is shown in the figure;

[0043] Figure 2 The structure diagram of the Bayesian neural network proxy model shown in the present application is shown in the figure;

[0044] Figure 3 The online correction process diagram of the performance degradation model parameter based on the sequential Bayesian method shown in the present application is shown in the figure;

[0045] Figure 4 The output interface diagram of the intelligent system shown in the present application is shown in the figure. DETAILED DESCRIPTION

[0046] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the specific embodiments of the present application will be described in detail below with reference to the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments.

[0047] Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor should be within the scope of protection of the present application.

[0048] In the following description, a lot of specific details are set forth in order to provide a thorough understanding of the present application, however, the present application can be practiced in other manners different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the present application, therefore, the present application is not limited to the specific embodiments disclosed below.

[0049] The embodiment takes the bottom pivot of a miter gate of a certain inland river shipping lock as an example, and details the complete implementation process of a real-time prediction and reliability evaluation method for the mushroom head wear of the lock bottom pivot.

[0050] With reference to Figure 1 The implementation steps of the method are as follows:

[0051] (1) Establish a finite element model of the lock to obtain simulation data of the wear mechanism

[0052] In the finite element software ANSYS, a three-dimensional refined finite element model including the miter gate leaf, top pivot and bottom pivot is established, and the material properties, contact parameters and boundary conditions of the model are set according to the design drawings and material manual; in order to ensure that the model can truly reflect the mechanical properties of the actual structure, the stress calculation results of the main beam and web of the finite element model can be compared with the stress monitoring data at the same location on site to cover the opening and closing operation conditions of the lock miter gate.

[0053] In the three-dimensional finite element model, by modifying the geometric shape of the contact surface of the mushroom head of the bottom pivot, 11 different wear depth levels from healthy to near failure are systematically simulated, specifically: 0mm (healthy), 2mm, 4mm, 6mm, 8mm, 10mm, 12mm, 14mm, 16mm, 18mm, 20mm (near failure), the simulation method adopts transient dynamics analysis, under each wear level, the working condition load of the upstream water level of 3m, 5m, 7m is applied respectively, and the transient dynamics simulation of the whole process of gate opening and closing is carried out, in the simulation process, by means of embedding user subprogram, the Archard wear model is used to calculate and accumulate the wear volume of the contact area in real time, and the Archard wear model is expressed as:

[0054]

[0055] Wherein, is the wear volume, is the dimensionless wear coefficient, is the normal load, is the relative sliding distance, is the material hardness.

[0056] Measuring points are set at key locations such as the door leaf panel, main beam, and bottom pivot base in the simulation model. Time-history data such as vibration acceleration and von Mises stress at these locations during the opening and closing process are extracted. To convert the simulation-obtained time-history data into a format suitable for machine learning, feature engineering is required. The feature extraction method is time-domain statistical analysis. For the extracted vibration acceleration and von Mises stress time-history data x at each measuring point, the root mean square (RMSE), peak value, and kurtosis are calculated. The RMSE reflects the energy magnitude of the signal, and the calculation formula is as follows:

[0057]

[0058] The peak value reflects the maximum impact level of the signal, and the calculation formula is:

[0059]

[0060] Kurtosis reflects the impulsiveness or outlier distribution of a signal, and is calculated using the following formula:

[0061]

[0062] in, Let be the sampled value of the signal at time point i, and N be the total number of sampling points. This represents the signal mean.

[0063] The calculation results of all simulation conditions (11 wear levels × 3 water level conditions) are summarized into a structured feature dataset, which is expressed in the form of a... The numerical matrix, where The total number of samples is 11×3, and the last column is the output target. The total dimension D of the input features is specifically 19. For example, assuming there are 3 key measuring points (door leaf panel, main beam, and bottom pivot base), and 3 statistical features (RMSE, Peak, Kurtosis) of vibration acceleration and von Mises stress are extracted for each measuring point, then each measuring point contributes 6 features, for a total of 18 features, plus 1 feature for the water level condition, i.e. ;

[0064] For example, the input features are represented as a D-dimensional feature vector. ,For example The output target is expressed as a scalar value. , representing the final cumulative wear depth obtained from the Arcard model simulation under this working condition.

[0065] Finally, the input feature sample data set is formed, and each sample contains a set of input features (such as the root mean square value, peak value, kurtosis, and water level of the vibration signal) and the corresponding output target (cumulative wear depth).

[0066] (2) Build a wear state proxy model

[0067] Referring to Figure 2 , a Bayesian neural network is selected as the proxy model, and a Bayesian neural network with two hidden layers and 64 neurons in each layer is constructed, the number of nodes in the input layer corresponds to the number of dynamic response features extracted in step (1) (for example, 19), and the output layer is a single node representing the cumulative wear amount.

[0068] The data set generated in step (1) is divided into a training set and a test set in a ratio of 8:2, and a deterministic approximation method of the variational inference algorithm is used to train and predict the Bayesian neural network, which converts the problem of solving the posterior of the hidden parameter into a variational optimization problem of maximizing the evidence lower bound ELBO, and finds a better posterior distribution solution through iteration to realize the rapid prediction of the wear state of the bottom pivot;

[0069] Exemplarily, the specific training method is: initializing the prior distribution of the hidden variable Z as a Gaussian distribution N(0, 1); selecting a variational approximation distribution as a mean field Gaussian distribution; using an Adam optimizer (learning rate 0.001) for iterative optimization, and in each iteration, the expected item of ELBO is estimated by Monte Carlo sampling (sampling number S=50) , and the Kullback-Leibler divergence item is calculated; train for 100 epochs until ELBO converges (change <1e-4); the formula of ELBO is:

[0070]

[0071] wherein, ELBO is the evidence lower bound, which is the objective function to be maximized in variational inference, is the expectation of the variational approximation distribution , is the finite element simulation data of the input features of the proxy model, Z is the hidden variable, i.e. the weight and bias model parameters of the Bayesian neural network, is the variational approximation distribution used to approximate the true posterior distribution , is the likelihood function of the finite element simulation data given the hidden variable Z, is the prior distribution of the hidden variable Z, is the variational approximation distribution With prior distribution of latent variables The Kullback-Leibler divergence between them.

[0072] After training, the surrogate model is expressed as a parameterized probabilistic model: given input features The output is the posterior predicted distribution of cumulative wear. , through from Sample to generate multiple predicted values And calculate its mean. and variance As a point estimate and uncertainty quantification, this surrogate model has the ability to quickly provide a probabilistic estimate of wear based on input vibration, stress and other data.

[0073] As an example, the testing method is as follows: input the input features of the test set into the trained Bayesian neural network, obtain the probability distribution of the prediction results through multiple forward propagations, and calculate the prediction mean as the point estimate.

[0074] (3) Establish a performance degradation assessment model

[0075] Gamma stochastic process was selected to measure wear depth To model, its probability density function for:

[0076]

[0077] in, In time The cumulative wear amount is The probability density, In time The cumulative wear over time, It is the shape parameter of the Gamma distribution, a function of time, representing the degree of cumulative wear. Let be the scale parameter of the Gamma distribution, reflecting the average magnitude of wear, and e be the base of the natural logarithm. It is the Gamma function.

[0078] It should be noted that, To initialize the model, prior judgments need to be constructed using historical observation data. Specifically, this involves reviewing the lock's maintenance log and extracting reliable measurement records from previous major overhauls or special inspections regarding the wear of the replenished mushroom head on the bottom pivot. For example, this embodiment collects two data points: a wear depth of 5mm measured during a major overhaul in 2015 (5 years after commissioning) and a wear depth of 12mm measured during a major overhaul in 2020 (10 years after commissioning). Using these two historical observation data points, the initial estimates of the model parameters are obtained through maximum likelihood estimation (MLE). .

[0079] Further, The calculation process is as follows:

[0080] First, the cumulative wear is converted into increments (Δwi) years, ; years, ); assuming , the parameters ; construct the log-likelihood function:

[0081]

[0082] where, is the wear rate parameter, reflecting the growth rate of wear degree per unit time; is the scale parameter, controlling the scale of wear distribution; is the duration of the i-th time period, is the wear increment in the i-th time period, is the Gamma function.

[0083] Use numerical optimization algorithms (such as L-BFGS-B) to maximize the function, maximize the above log-likelihood function under the constraint , Set the initial guess value (such as ), substitute the increment data , iterate the gradient and update the parameter value, repeat until convergence (such as relative change <1×10⁻ 6 ), and finally obtain the parameter estimation result: , so , this constitutes a priori judgment of the long-term wear trend of the mushroom head.

[0084] (4) Online correction and dynamic prediction

[0085] During the actual operation of the ship lock, the acceleration sensor arranged on the bottom pivot base is used to collect vibration signals in real time. At time k (for example, after 2000 cycles of operation), the system automatically collects vibration data for a complete opening and closing cycle, that is, the entire operation process of the miter gate from the completely closed state, through the opening process, maintaining the open state, the closing process, and finally returning to the completely closed state. The total duration of the example is about 10-15 minutes, of which the dynamic opening and closing process accounts for about 3 minutes, which is the key period for data collection. The system automatically processes the 3-minute vibration time series data extracts the same statistical features as step (1), including root mean square, peak value, and kurtosis, and combines the current water level working condition to form a feature vector .

[0086] The vector is input into the trained Bayesian neural network agent model in step (2) to obtain a probabilistic estimate of the current cumulative wear, specifically: the feature vector is normalized according to the standardization parameters during training; M times (such as M = 100) are sampled from the trained variational posterior distribution to obtain network weight samples ; for each weight sample , the corresponding predicted value is calculated; the mean and variance of the prediction distribution are calculated, for example .

[0087] Referring to Figure 3 , the wear estimate value obtained in the last step is taken as the latest observation evidence in the sequential Bayesian update formula , and the particle filtering algorithm is used to iteratively update the parameters of the Gamma process, and the specific iterative update method is as follows:

[0088] (a) initialization / prediction: N particles (parameter samples) are sampled from the parameter posterior distribution at the previous time (k-1) to generate , for example N = 1000, since wear is a slow-changing process, the parameters remain unchanged in the prediction step, i.e. ;

[0089] (b) weight calculation: according to the latest observation value and its variance , the importance weight of each particle is calculated, and the weight is proportional to the likelihood (likelihood) of the model parameter represented by the particle generating the observation value ; assuming that the observation error is Gaussian distribution, then the likelihood function is:

[0090]

[0091] where is the wear mean predicted by the particle ;

[0092] (c) resampling: according to the calculated weight, the particle set is resampled; particles with high weight have a higher probability of being copied, and particles with low weight may be discarded, and a new set of equally weighted particles is obtained after resampling, which approximately represents the updated parameter posterior distribution , such as Figure 3As shown, this process corrects the original predicted trajectory (dashed line) to a new predicted trajectory (solid line) that is closer to the current actual state.

[0093] Assuming a wear failure threshold It is 20mm, based on the corrected parameters. posterior distribution Recalculate the reliability of the bottom pivot at the current moment. The formula is:

[0094]

[0095] Predict the probability distribution of remaining useful life (RUL), where RUL is the first time the wear increment reaches the remaining wear amount. The time required for each updated particle Increment at any future time Internal wear increment It follows a Gamma distribution with shape parameters as follows: The scale parameter is The cumulative distribution function (CDF) of RUL is:

[0096]

[0097] The final RUL probability density function (PDF) can be obtained by taking the Monte Carlo average of the RUL distributions calculated for all N particles.

[0098] Reference Figure 4 The intelligent monitoring platform's visual interface displays the updated results in real time, such as... Figure 4 As shown, the current reliability is 99.6%, and the probability density function of RUL shows that its expected value is 3.8 years. Since the 5th percentile of RUL (e.g., 3.1 years) is less than the preset next major overhaul cycle (4 years), the system automatically triggers a yellow warning and pushes a report to the operation and maintenance management personnel, suggesting that they conduct a key inspection of the bottom pivot during the next maintenance window, providing decision support for them to formulate a scientific predictive maintenance strategy.

[0099] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for real-time prediction of lock bottom pivot mushroom head wear and reliability assessment, characterized by, The method comprises the steps of: The finite element simulation model adopts Lagrange contact algorithm to realize friction simulation of the mushroom head, and calculates the wear amount of the mushroom head of the bottom pivot by Archard wear model, and the Archard wear model is expressed as: A Bayesian neural network proxy model, i.e., a mushroom head wear state proxy model, is constructed and trained based on the Bayesian neural network prediction algorithm by using the obtained finite element simulation data, and the mushroom head wear state proxy model is used to efficiently predict the wear amount of the mushroom head according to the structural dynamic response simulation data; wherein, is the wear volume, is the dimensionless wear coefficient, is the normal load, is the relative sliding distance, is the material hardness; The input features of the Bayesian neural network proxy model are selected from the feature combinations of one or more physical quantities of stress, strain, vibration response, acoustic emission signal, driving mechanism motor current and gate operation water level of the gate leaf and the supported rotating part, and the output feature is the cumulative wear amount of the mushroom head; A mushroom head performance degradation evaluation model is established based on Gamma random process theory to describe the long-term random process of wear evolution over time, and historical data are used to initialize and estimate the model parameters; In combination with the real-time monitored structural dynamic response data, the current wear amount is output by the mushroom head wear state proxy model, and then the wear amount is taken as new observation data, and the parameters of the mushroom head performance degradation evaluation model are corrected in real time by using the sequential Bayesian method, so that the reliability and residual life prediction result of the mushroom head wear failure are dynamically updated; The cumulative wear amount of the mushroom head performance degradation evaluation model The probability density function of the cumulative wear amount is: wherein, is the probability density of the cumulative wear amount at time is the cumulative wear amount at time is the cumulative wear amount at time is the cumulative wear amount at time is the cumulative wear amount at time is the shape parameter of the Gamma distribution, is a function of time, and indicates the degree of cumulative wear, is the scale parameter of the Gamma distribution, reflects the average size of the wear amount, and e is the base of the natural logarithm, is the Gamma function; initial parameters of the mushroom head performance degradation evaluation model The estimation is performed by a maximum likelihood estimation method whose objective function is in, Let be the likelihood function. Represents the parameters of the Gamma stochastic process model, i.e. and , Historical observation data used for parameter estimation In time Observed cumulative wear For given parameters Cumulative wear during the Gamma stochastic process The probability density function; The Bayesian neural network proxy model is trained and predicted by using the deterministic approximation method of variational inference algorithm, the solving of the hidden parameter posterior problem is converted into the variational optimization problem of maximizing the evidence lower bound ELBO, and the optimal posterior distribution solution is found by iteration, so that the wear state of the bottom pivot is quickly predicted, and the formula of the ELBO is: The sequence Bayesian method corrects parameters of the mushroom head performance degradation evaluation model in real time according to Bayesian theorem Iterative updating is performed, and a mathematical expression formula is: in, In order to observe real-time monitoring data After that, parameters The posterior probability distribution, For given parameters In the case of the latest monitoring data The likelihood function, Let be the posterior probability distribution of the previous time step, and be the updated prior probability distribution. Indicates proportionality. These are the parameters for the Gamma stochastic process model.

2. The real-time prediction of lock bottom pivot mushroom head wear and reliability assessment method according to claim 1, characterized in that, After the reliability and residual life prediction result of the mushroom head wear failure are dynamically updated, a set of predictive maintenance strategy of the mushroom head is formulated and implemented, which specifically comprises the following steps: in, Let ELBO be the lower bound of evidence, which is the objective function to be maximized in variational inference. For variational approximation distribution Seeking expectations, The finite element simulation data that serves as input features to the surrogate model, where Z represents latent variables, i.e., the weights and biases of the Bayesian neural network model parameters. For approximating the true posterior distribution The variational approximation distribution, Finite element simulation data given the latent variable Z The likelihood function, Let Z be the prior distribution of the latent variable. For variational approximation distribution With prior distribution of latent variables The Kullback-Leibler divergence between them.

3. The real-time prediction of lock bottom pivot mushroom head wear and reliability assessment method according to claim 1, characterized in that, The step further comprises formulating and executing the maintenance strategy, wherein: Cumulative distribution function of remaining life: based on the real-time updated mushroom head performance degradation assessment model, a predefined wear-out failure threshold is set , the cumulative distribution function of the remaining life after the current time is calculated , which represents the probability that the cumulative wear of the bottom pivot first reaches or exceeds the failure threshold within the future time ;​ establishing a predictive maintenance strategy objective function: establishing a function with the goal of minimizing the long-term expected total cost which is used to evaluate the economic efficiency of performing preventive maintenance at a future maintenance opportunity in particular in the form of: wherein, Cpmis the cost of performing a preventive maintenance, Cpmis the cost of performing a preventive maintenance, P(t) is the probability of failure before time t, P(t) is the probability of failure before time t, P(t) is the probability of failure before time t, P(t) is the probability of failure before time t. Determining optimal preventive maintenance timing: determining the optimal preventive maintenance timing by solving an optimization problem i.e. finding the maintenance time point that minimizes the objective function defined in the previous step :​ The optimal preventive maintenance timing A cost-benefit analysis based optimal maintenance decision window is provided.

4. The real-time prediction of lock bottom pivot mushroom head wear and reliability assessment method according to claim 3, characterized in that, ​ based on the calculated optimal preventive maintenance timing in combination with the current bottom pivot operating time and status, a maintenance recommendation is generated; When the actual running time is close to the system triggers a maintenance alarm, instructs relevant personnel to perform preventive maintenance, and feeds back the maintenance data to the historical database after the maintenance is completed, for the initialization and update of the next round of model, forming a continuous optimization closed-loop management process.

Images