Collaborative self-renewal life evaluation method for two-phase degradation of coating structure

Abstract

The invention provides a collaborative self-updating life evaluation method for coating structure double-phase degradation. The method comprises the following steps: 1, coating-structure double-phase degradation modeling; 2, degeneration parameter collaborative self-updating; and 3, predicting the failure probability and the residual life. The method has the advantages that the layered random process model is established, the dynamic evolution process of coating degradation and structure corrosion is represented, the interaction between the two is considered, the coating-structure failure probability is predicted, and the service life evaluation accuracy is improved; according to the method, a Bayesian dynamic updating method can be adopted to carry out collaborative correction on coating and structure degradation parameters, so that the influence of uncertain factors such as environment, material characteristics and operation conditions is effectively dealt with, and the model adaptability and real-time updating capability are improved; the method is scientific, is particularly suitable for predicting the service life of the metal structure in complex service environments such as ocean, can provide a scientific basis for maintenance of a corrosion protection system, and ensures long-term safe service of the structure.

Description

TECHNICAL FIELD

[0001] The application provides a synergistic self-updating life evaluation method for two-phase degradation of a coating structure, that is, a structure reliability and life evaluation method based on two-stage degradation process modeling and Bayesian parameter updating. In view of the two-phase degradation characteristics of coating degradation and structure corrosion in a corrosion protection system, a dependent process model is established to represent the coating degradation and structure degradation trajectories, the degradation parameter uncertainty is handled through synergistic self-updating fusion of field detection data, the failure probability evaluation and residual life prediction of a structure containing a corrosion protection coating are realized, and theoretical and technical support is provided for improving the safety of structures in service in complex environments such as the sea. The application is applicable to failure probability evaluation and residual life prediction of a metal structure containing a corrosion protection coating under the condition that the coating degradation affects the substrate degradation. BACKGROUND

[0002] Metal structures such as marine equipment and offshore bridges are subjected to the coupling of multiple fields such as high salt mist corrosion, dry-wet alternating, and microbial attachment during service, resulting in the common existence of interactive degradation phenomena such as powdering and peeling of protective coatings and expansion of pitting corrosion of the substrate. The coating-substrate two-phase degradation process causes the performance degradation of the structure to present obvious two-stage characteristics: a slow performance degradation stage in the coating protection period and an accelerated failure stage in the substrate corrosion period. Engineering practice shows that critical degradation behaviors such as chloride ion penetration caused by coating microcracks and sudden increase of corrosion current density caused by rupture of the passivation film of the substrate can cause the residual life prediction error of the structure to exceed 35%, seriously threatening the service safety of major equipment.

[0003] However, the current life evaluation methods for corrosion protection structures have significant limitations: on the one hand, traditional methods regard coating failure and substrate corrosion as independent processes (such as the staged evaluation mode in GB / T 30790 standard), do not establish a dynamic correlation model of two-stage degradation parameters, and cannot capture the accelerated corrosion characteristics of the structure caused by critical coating failure; on the other hand, existing models mostly use fixed parameter settings (such as preset coating degradation rate in NORSOK M-501 standard), lack dynamic fusion capability of field detection data, and cause parameter estimation error to be amplified with service time.

[0004] To solve the above problems, the application provides a synergistic self-updating life evaluation method for two-phase degradation of a coating structure. The method establishes a dependent process model to represent the coating degradation and structure degradation trajectories in view of the two-phase degradation characteristics of coating degradation and structure corrosion in a corrosion protection system, handles the degradation parameter uncertainty through synergistic self-updating fusion of field detection data, and realizes failure probability evaluation and residual life prediction of a structure containing a corrosion protection coating. SUMMARY

[0005] (1) The purpose of the present application: to cope with the failure probability evaluation and residual life prediction of metal structures with corrosion protection coating under the influence of coating degradation on substrate degradation scenarios, aiming at the dual degradation characteristics of coating degradation and structure corrosion in the corrosion protection system, a dependent process model is established to characterize the coating degradation and structure degradation trajectory, the degradation parameter uncertainty is processed through the cooperative self-updating of the fusion of field detection data, the failure probability evaluation and residual life prediction of the structure with corrosion protection coating are realized, and the theoretical and technical support is provided for improving the safety of structure service in complex environment such as sea.

[0006] (2) Technical scheme:

[0007] The present application provides a cooperative self-updating life evaluation method for dual-phase degradation of coating structure, that is, a structure reliability and life evaluation method based on two-stage degradation process modeling and Bayesian parameter updating, which is realized through the following three steps:

[0008] Step one: coating-structure dual-phase degradation modeling

[0009] In the corrosion protection system, the degradation of coating not only directly affects its protection function, but also accelerates the corrosion of the substrate structure, leading to the decline of structure safety. Therefore, step one adopts a dependent process to characterize the interaction between coating and structure degradation.

[0010] The degradation of coating mainly manifests as the gradual reduction of coating thickness, the reduction of adhesion, cracking or peeling, etc. The degradation of coating is a gradual process, which is usually affected by environmental factors (such as humidity, temperature, salt spray, etc.), and presents a monotonically increasing trend. Therefore, the coating degradation can be modeled by a monotonically continuous increasing Gamma process:

[0011] ΔX c (t)~Gamma(α c Δt,β c )(1)

[0012] Where X c (t) is the degradation amount of coating at time t; ΔX c (t) is the degradation increment of X c (t) in Δt period; α c is the shape parameter of coating degradation process, reflecting the degradation rate of coating per unit time; β c is the scale parameter, describing the discreteness and expansibility of coating degradation process.

[0013] The degradation process of the structure also presents a monotonic increasing trend, but unlike the coating degradation, the rate of structure degradation is not only affected by its own material and environment, but also by the coating degradation. With the gradual degradation of the coating, the substrate is gradually exposed to the corrosion environment, and the degradation rate significantly accelerates. Therefore, the rate of structure degradation is closely related to the state of coating degradation. To describe this interaction, we use a state-dependent Gamma process, that is, the structure degradation rate is a function of the coating degradation state:

[0014] ΔX s (t)~Gamma(a s f(X c (t))Δt,β s )(2)

[0015] Where X s (t) is the degradation amount of the structure at time t; ΔX s (t) is the degradation increment of X s (t) in the time period Δt; a c is the shape parameter of the structure degradation process, reflecting the degradation rate of the structure per unit time; f(X c (t)) is the influence function of coating degradation on structure degradation rate, which represents the relationship between structure degradation rate and coating degradation degree, and the common models are linear function or exponential function; β s is the scale parameter, describing the discreteness and expansibility of the structure degradation process.

[0016] Step two: Degradation parameter collaborative self-update

[0017] Because the degradation process of the corrosion protection system is affected by many uncertain factors, including environmental conditions, material properties, operating conditions, and detection errors. These factors make the degradation parameters (such as coating degradation rate, structure corrosion rate, etc.) have high uncertainty. Therefore, before failure probability prediction and remaining life assessment, the coating degradation parameter and structure degradation parameter estimate value need to be collaboratively updated, so as to be able to adjust the model in real time under the condition of continuous acquisition of field detection data, and provide more accurate prediction.

[0018] The Bayesian dynamic updating method provides a method for updating the uncertainty of degradation parameters by continuously integrating field monitoring data. Specifically, the Bayesian method uses prior information and newly acquired detection data to update the parameters in the degradation model, thereby improving the accuracy of parameter estimation.

[0019] In the case of insufficient understanding of coating and structure degradation processes, the prior distribution of the shape parameter and scale parameter of the degradation can be described using a normal distribution, that is, Where μ and σ are the mean and standard deviation of the normal distribution, respectively. an initial mean of a coating degradation shape parameter; an initial variance of a coating degradation shape parameter; an initial mean of a coating degradation scale parameter; an initial variance of a coating degradation scale parameter; an initial mean of a structure degradation shape parameter; an initial variance of a structure degradation shape parameter; an initial mean of a structure degradation scale parameter; an initial variance of a structure degradation scale parameter.

[0020] Assume at time t k the kth field data is detected: coating degradation value x c (t k ) and structure degradation value x s (t k ), the error of measurement data obeys normal distribution, i.e. wherein, denotes the variance of measurement error. Therefore:

[0021]

[0022] wherein, P(X c (t k )|α c ) denotes the conditional probability that the coating degradation value is x c (t c ) under the condition that the coating degradation shape parameter is α k ; P(X c (t k )|β c ) denotes the conditional probability that the coating degradation value is x c (t c ) under the condition that the coating degradation scale parameter is β k ; P(X s (t k )|α s ) denotes the conditional probability that the structure degradation value is x s (t s ) under the condition that the structure degradation shape parameter is α k ; P(X s (t k )|β s ) denotes the conditional probability that the structure degradation value is x s (t s ) under the condition that the structure degradation scale parameter is β k .

[0023] According to Bayesian derivation:

[0024]

[0025] where P(a c | x c (t k )) is the conditional probability of the coating degradation shape parameter given that the degradation value of the coating is x c (t k ); P(b c | x c (t k )) is the conditional probability of the coating degradation scale parameter given that the degradation value of the coating is x c (t k ); P(a s | x s (t k )) is the conditional probability of the structure degradation shape parameter given that the degradation value of the structure is x s (t k ); P(b s | x s (t k )) is the conditional probability of the structure degradation scale parameter given that the degradation value of the structure is x s (t k ); P(a c ) is the probability of the coating degradation shape parameter; P(b c ) is the probability of the coating degradation scale parameter; P(a s ) is the probability of the structure degradation shape parameter; and P(b s ) is the probability of the structure degradation scale parameter.

[0026] With Equations (3)-(4) and the prior distributions of the degradation shape and scale parameters, the posterior distributions of the degradation shape and scale parameters are approximately normal distributions. The collaborative updating formulas of the degradation shape and scale parameters are approximately as follows:

[0027]

[0028] where is the mean of the coating degradation shape parameter after the k-1th detection; is the variance of the coating degradation shape parameter after the k-1th detection; is the mean of the coating degradation shape parameter after the kth detection; is the variance of the coating degradation shape parameter after the kth detection; is the mean of the coating degradation scale parameter after the k-1th detection; and represents the mean of the coating degradation scale parameter after the kth inspection; represents the mean of the coating degradation scale parameter after the kth inspection; represents the mean of the coating degradation scale parameter after the kth inspection; represents the mean of the structure degradation shape parameter after the k-1th inspection; represents the mean of the structure degradation shape parameter after the k-1th inspection; represents the mean of the structure degradation shape parameter after the k-1th inspection; represents the mean of the structure degradation shape parameter after the k-1th inspection; represents the mean of the structure degradation scale parameter after the k-1th inspection; represents the mean of the structure degradation scale parameter after the k-1th inspection; represents the mean of the structure degradation scale parameter after the k-1th inspection; represents the mean of the structure degradation scale parameter after the k-1th inspection.

[0029] After obtaining the first field data of the coating degradation value x c (t1) and the structure degradation value x s (t1), the initial values of the degradation shape parameter and the scale parameter are substituted into equation (5) to obtain the updated values of the degradation shape parameter and the scale parameter after Bayesian updating. In the same way, the degradation parameters can be continuously updated by fusing field monitoring data, thereby improving the accuracy of parameter estimation.

[0030] Step three: failure probability and residual life prediction

[0031] Assuming that the critical degradation value of the structure failure is L s , the failure probability of the structure can be obtained by the Gamma process:

[0032]

[0033] where P(T s ≤t) represents the failure probability of the structure; γ(·) is the lower incomplete Gamma function; and Γ(·) is the Gamma function.

[0034] According to the mathematical expectation of the Gamma process, the expected residual life of the structure satisfies:

[0035]

[0036] where E(T s ) is the expected residual life of the structure, which needs to be obtained by Monte Carlo simulation after substituting the degradation data and parameter values obtained in step two.

[0037] Through the above steps, the coating degradation and structure degradation trajectory modeling, the degradation parameter Bayesian dynamic updating, and the failure probability evaluation and residual life prediction of the coating and structure are realized, and the problem of how to scientifically evaluate the failure probability and predict the residual life of the metal structure with corrosion protection coating under the influence of coating degradation on the base degradation scene is solved.

[0038] (3) Advantages and effects:

[0039] 1. The present application can predict the coating-structure failure probability by establishing a dependent process model, representing the dynamic evolution process of coating degradation and structure corrosion, and considering the interaction between the two, and improving the accuracy of life evaluation;

[0040] 2. The present application can use the Bayesian dynamic updating method to correct the coating and structure degradation parameters, effectively cope with the influence of environmental, material properties and operating conditions and other uncertain factors, and improve the model adaptability and real-time updating ability;

[0041] 3. The method described in the present application is scientific, especially suitable for life prediction of metal structures in complex service environments such as the sea, and can provide a scientific basis for the maintenance of corrosion protection systems and ensure the long-term safe service of the structure. BRIEF DESCRIPTION OF DRAWINGS

[0042] Figure 1 The flow chart of the method described in the present application.

[0043] Figure 2 The coating and structure degradation curves of the support beam.

[0044] Figure 3 The failure probability curve of the support beam. DETAILED DESCRIPTION

[0045] The present application will be further described below with reference to examples.

[0046] A certain type of ship deck support beam serves in the complex marine environment for a long time, and the external coating is used to prevent the corrosion of corrosive seawater and humidity on the support beam. However, as the service time increases, the coating gradually ages and peels off, causing the support beam base metal to be directly exposed to the harsh environment, accelerating the corrosion process of the support beam, and eventually leading to the deck fracture. Therefore, modeling the two-phase degradation of the coating and the support beam structure, and life prediction based on the Bayesian updating failure probability prediction method is of great importance for the maintenance of the ship deck.

[0047] Table 1 and Figure 2 shows the coating and structure degradation data of a certain support beam in the first year of corrosion environment. From Figure 2As can be seen from the figure, the degradation amount of the coating and the structure monotonically increases with time, without decreasing or oscillating trend, while the increment has certain volatility, which is more consistent with the monotonic increase of the Gamma process than the commonly used Wiener process in engineering, so the Gamma process can be used to model the degradation trajectory of the coating and the structure. It is also noted that with the degradation of the coating, the degradation rate of the structure presents an exponential increase, so it can be considered that the influence function of the coating degradation on the structure degradation rate is an exponential function.

[0048] Table 1 Coating and structure degradation data of a certain support beam in the first year under corrosion environment

[0049]

[0050]

[0051] The present application provides a kind of synergistic self-renewal life evaluation method for coating structure two-phase degradation, i.e. a kind of structure reliability and life evaluation method based on two-stage degradation process modeling and Bayes parameter update, see Figure 1 As shown in the figure, the method is realized by the following 3 steps:

[0052] Step one: coating-structure two-phase degradation modeling

[0053] Gamma process is used to model the coating and structure degradation of the support beam:

[0054] ΔX c (t)~Gamma(α c Δt,β c )

[0055] ΔX s (t)~Gamma(α s f(X c (t))Δt,β s )

[0056] f(X c (t))=exp(λX c (t))

[0057] Wherein, exp(·) is natural exponential function, and λ is coupling parameter.

[0058] The method of matrix estimation can be used to quickly estimate the shape parameter and scale parameter of Gamma process, i.e.:

[0059]

[0060] Wherein, sample refers to degradation increment, which is obtained by subtracting the degradation amount of two consecutive weeks.

[0061] The initial estimates of the shape and scale parameters of the coating and structure degradation are obtained as follows:

[0062]

[0063] The coupling parameter can be estimated by using the maximum likelihood estimation, i.e.,

[0064]

[0065] where ln(·) is the natural logarithm function.

[0066] Substituting the data in Table 1 and the initial estimates of the shape and scale parameters into the above equation, the estimate of the coupling parameter is obtained as follows: λ = 0.015.

[0067] Step 2: Degradation parameter collaborative self-updating

[0068] Table 2 shows the newly added coating and structure degradation data of the support beam in the corrosion environment. First, we assume that the prior distribution of the shape parameter and the scale parameter is α c ~ N(0.62, 0.2 2 ), β c ~ N(4.41, 0.5 2 ), α s ~ N(0.24, 0.08 2 ), β s ~ N(3.28, 0.5 2 ), and the measurement error ε ~ N(0, 1 2 ). Then, the mean values of the shape and scale parameters are dynamically updated using the Bayesian method with the new data in Table 2, and the updated results are shown in Table 3.

[0069] Table 2 Newly added coating and structure degradation data of a support beam in a corrosion environment

[0070] Time (weeks) Coating degradation (microns) Structure degradation (microns) 49 134 38 50 139 40 51 144 43 52 149 47

[0071] Table 3 Mean value update results of coating and structure degradation parameters

[0072]

[0073] Step 3: Failure probability and remaining life prediction

[0074] Given that the critical degradation value of the structure failure of the support beam is L s = 10000 (microns), the structure failure probability curve can be obtained by the Gamma process as shown in Figure 3 . By substituting the parameter values into equation (7) and performing Monte Carlo simulation, the expected remaining life of the structure converges to: E(T s)=44(weeks).

[0075] The results show that the coating degradation and structure degradation trajectory modeling, degradation parameter Bayesian collaborative self-updating, and failure probability evaluation and residual life prediction of the coating and structure can be realized by using the application, the problem of how to scientifically evaluate the failure probability and predict the residual life of the metal structure with corrosion protection coating under the influence of coating degradation on the base degradation scene is solved, and the expected purpose is achieved.

[0076] In conclusion, the application provides a collaborative self-updating life evaluation method for two-phase degradation of coating structure, that is, a structure reliability and life evaluation method based on two-stage degradation process modeling and Bayesian parameter updating. According to the two-phase degradation characteristics of coating degradation and structure corrosion in the corrosion protection system, a dependent process model is established to characterize the coating degradation and structure degradation trajectory, the degradation parameter uncertainty is collaboratively and self-updatingly processed by fusing the field detection data, the failure probability evaluation and residual life prediction of the structure with corrosion protection coating are realized, and theoretical and technical support is provided for improving the safety of structure service in complex environments such as the sea.

Claims

1. A method for synergistic self-renewal life assessment against bi- phase degradation of a coating structure, characterized by, The method comprises the following steps: Step one: coating-structure degradation modeling A dependent process is used to characterize the interaction between coating and structure degradation; Step two: degradation parameter collaborative updating Before failure probability prediction and residual life assessment, the coating degradation parameter and structure degradation parameter estimates need to be collaboratively updated. Bayesian method is used to update the parameters in the degradation model using prior information and newly acquired detection data to improve the accuracy of parameter estimation; Step three: failure probability and residual life prediction Failure probability assessment and residual life prediction of the structure with corrosion protection coating.

2. A method for synergistic self-updating life assessment against dual-phase degradation of coating structure according to claim 1, characterized in that: In step one, coating degradation is modeled by a monotonically increasing Gamma process: ΔX c (t) ~ Gamma(a c Δt,β c )(1) wherein X c (t) is the degradation amount of the coating at time t; ΔX c (t) is the degradation amount of the coating at time t; ΔX c (t) is the degradation increment of X c is a shape parameter of the coating degradation process, reflecting the degradation rate of the coating per unit time; β c is a scale parameter, describing the discreteness and expansibility of the coating degradation process.

3. A synergistic self-renewal life assessment method for coating structure dual-phase degradation according to claim 1 or 2, characterized in that: The structure degradation process also shows a monotonically increasing trend, and the rate of structure degradation is closely related to the state of coating degradation; a state-dependent Gamma process is used, that is, the rate of structure degradation is a function of the state of coating degradation: ΔX s (t) ~ Gamma(a s f(X c (t)) Δt, β s )(2) where X s (t) is the degradation amount of the structure at time t; ΔX s (t) is the degradation amount of the structure at time t; ΔX s (t) is the degradation increment of the structure in the time period of Δt; α c is the shape parameter of the structure degradation process, reflecting the degradation rate of the structure per unit time; f(X c (t)) is the influence function of coating degradation on the structure degradation rate, which represents the relationship between the structure degradation rate and the coating degradation degree, and the model is a linear function or an exponential function; β s is the scale parameter, describing the discreteness and expansibility of the structure degradation process.

4. The method for synergistic self-healing life assessment of coating structure dual-phase degradation according to claim 1, characterized in that: In step two, a normal distribution is used to describe the prior distribution of the degradation shape parameter and scale parameter, i.e. where, represents the initial mean of the coating degradation shape parameter; represents the initial variance of the coating degradation shape parameter; represents the initial mean of the coating degradation scale parameter; represents the initial variance of the coating degradation scale parameter; represents the initial mean of the structure degradation shape parameter; represents the initial variance of the structure degradation shape parameter; represents the initial mean of the structure degradation scale parameter; represents the initial variance of the structure degradation scale parameter.

5. A method for synergistic self-healing life assessment against biodegradation of coating structure according to claim 4, characterized in that: is set at time t k The kth field data is detected: coating degradation value x c (t k ) and structural degradation value x s (t k ), the measurement error is subject to normal distribution, that is Where, Indicates the variance of the measurement error; therefore: where P(X c (t k )|α c ) represents the conditional probability of the degradation value of the coating being x c (t c ) under the condition that the shape parameter of the coating degradation is α k ; P(X c (t k )|β c ) represents the conditional probability of the degradation value of the coating being x c (t c ) under the condition that the scale parameter of the coating degradation is β k ; P(X s (t k )|α s ) represents the conditional probability of the degradation value of the structure being x s (t s ) under the condition that the shape parameter of the structure degradation is α k ; and P(X s (t k )|β s ) represents the conditional probability of the degradation value of the structure being x s (t s ) under the condition that the scale parameter of the structure degradation is β k .

6. A method for synergistic self-healing life assessment against biodegradation of coating structure according to claim 5, characterized in that: According to Bayesian derivation: Wherein, P(α) c |x c (t k )) indicates that the detected degradation value of the coating is x. c (t k Under the condition of ), the conditional probability of coating degradation shape parameters; P(β) c |x c (t k )) indicates that the detected degradation value of the coating is x. c (t k Under the condition of ), the conditional probability of the coating degradation scale parameter; P(α) s |x s (t k )) indicates that when a degradation value of x is detected in the structure. s (t k Under the condition of ), the conditional probability of the structural degradation shape parameter; P(β) s |x s (t k )) indicates that when a degradation value of x is detected in the structure. s (t k Under the condition of ), the conditional probability of the structural degradation scale parameter; P(α) c P(β) represents the probability of coating degradation shape parameters; c P(α) represents the probability of the coating degradation scale parameter; s P(β) represents the probability of structural degradation shape parameters; s The probability of the structural degradation scale parameter.

7. A method for synergistic self-healing life assessment against biodegradation of coating structure according to claim 6, characterized in that: Combining equations (3)-(4) and the prior distribution of the degradation shape parameter and the scale parameter, the posterior distribution of the degradation shape parameter and the scale parameter is approximately normally distributed; the collaborative updating formula of the degradation shape parameter and the scale parameter is approximately as follows: wherein, represents the mean of the coating degradation shape parameter after the k-1th inspection; represents the variance of the coating degradation shape parameter after the k-1th inspection; represents the mean of the coating degradation shape parameter after the kth inspection; represents the variance of the coating degradation shape parameter after the kth inspection; represents the mean of the coating degradation scale parameter after the k-1th inspection; represents the variance of the coating degradation scale parameter after the k-1th inspection; represents the mean of the coating degradation scale parameter after the kth inspection; represents the variance of the coating degradation scale parameter after the kth inspection; represents the mean of the structure degradation shape parameter after the k-1th inspection; represents the variance of the structure degradation shape parameter after the k-1th inspection; represents the mean of the structure degradation shape parameter after the kth inspection; represents the variance of the structure degradation shape parameter after the kth inspection; represents the mean of the structure degradation scale parameter after the k-1th inspection; represents the variance of the structure degradation scale parameter after the k-1th inspection; represents the mean of the structure degradation scale parameter after the kth inspection; represents the variance of the structure degradation scale parameter after the kth inspection; At the time t1, the first on-site data is obtained: the coating degradation value x c (t1) and the structure degradation value x s After (t1), the initial values of the degraded shape parameters and scale parameters are substituted into equation (5) to obtain the updated values of the Bayesian updated degradation shape parameters and scale parameters.

8. A method for synergistic self-healing life assessment against biodegradation of coating structure according to claim 1, characterized in that: In step three, let the critical degradation value for structural failure be L s Then the failure probability of the structure is obtained by the Gamma process: where P(T s ≤t) denotes the structural failure probability; γ(·) is the lower incomplete Gamma function; Γ(·) is the Gamma function; According to the mathematical expectation of the Gamma process, the expected residual life of the structure satisfies: where E(T s ) is the expected residual lifetime of the structure, which needs to be obtained by Monte Carlo simulation after the degradation data and parameter values obtained in Step 2 are substituted.

Images