Polarization potential monitoring signal fused subway steel fiber concrete structure corrosion current density time sequence prediction method

Abstract

The invention discloses a metro steel fiber concrete structure corrosion current density time sequence prediction method fused with polarization potential monitoring signals, and the method comprises the steps: measuring corrosion current density time sequence data in a small period, taking the time sequence data in the small period as training data, constructing a corrosion current density time sequence prediction method based on a long short-term memory network (LSTM), and predicting the corrosion current density time sequence of a metro steel fiber concrete structure. According to the method, the development condition of the stray current corrosion of the steel fiber reinforced concrete structure can be accurately mastered through the corrosion current density time sequence prediction, so that the stray current corrosion early warning is realized, and the early warning of the stray current corrosion of the steel fiber reinforced concrete structure is realized. And an accurate basis is provided for carrying out stray current corrosion protection in time.

Description

Technical Field

[0001] This invention relates to a method for predicting corrosion current density, and more particularly to a time-series prediction method for corrosion current density of steel fiber reinforced concrete structures in subways that integrates polarization potential monitoring signals. Background Technology

[0002] Steel fiber reinforced concrete (SFRC) is a composite material formed by uniformly incorporating short, fine steel fibers into ordinary concrete. It effectively inhibits cracking, toughens the concrete, and significantly improves its tensile strength, flexural toughness, impact resistance, and fatigue resistance. In subway shield tunnels, SFRC is increasingly being used in precast segments. Compared to traditional reinforced concrete segments, SFRC segments utilize the bridging effect of fibers to inhibit crack propagation, enhancing the overall integrity, crack resistance, and durability of the segments. They are particularly well-suited to cope with the complex stresses, ground deformation, and potential localized impacts encountered during tunnel operation. Furthermore, the application of SFRC simplifies segment reinforcement (and can even partially replace steel mesh), optimizes the production process, improves construction efficiency, and facilitates segment thinning, demonstrating significant comprehensive technical and economic value.

[0003] During their service life, subway shield tunnels are significantly affected by stray current corrosion. The static pressure caused by the accumulation of corrosion products can severely impact the structural strength and service life of steel fiber reinforced concrete (SFC) segments. Therefore, the application of SFC in subway shield tunnels must also consider stray current corrosion protection. Given the numerous influencing factors of stray current corrosion, exploring the temporal patterns of stray current corrosion current density is helpful for early warning of stray current corrosion. This invention provides a temporal prediction method for corrosion current density of subway SFC structures that integrates polarization potential monitoring signals. By using the variation patterns of stray current corrosion current density within a short period obtained in a laboratory environment, the method realizes the variation patterns of corrosion current density within a long period in a real environment and compares them with the dangerous threshold of corrosion current density, thereby achieving early warning of stray current corrosion in SFC structures. Summary of the Invention

[0004] To achieve the above objectives, this invention provides a time-series prediction method for corrosion current density in subway steel fiber reinforced concrete structures that integrates polarization potential monitoring signals. This method aims to predict the time-series development of stray current corrosion in subway steel fiber reinforced concrete structures, providing effective support for early warning of stray current corrosion. The technical solution implemented by this invention is as follows:

[0005] A time-series prediction method for corrosion current density of steel fiber reinforced concrete structures in subways, incorporating polarization potential monitoring signals, includes the following steps:

[0006] Step 1: Set the subway's operating cycle duration to small cycles such as t1, t2, t3, ..., ti, ..., tn. Measure the corrosion current density time series data for the small cycles t1 to tn as icorr(t1), icorr(t2), icorr(t3), ..., icorr(ti), ..., icorr(tn). Measure the polarization potential time series data for the small cycles t1 to tn as P(t1), P(t2), P(t3), ..., P(ti), ..., P(tn). The calculation formula for the normalization processing method of the corrosion current density time series data and the polarization potential time series data is as follows:

[0007]

[0008] P max =max{P(t1),P(t2),P(t3),...,P(t i ),P(t i+1 ),...,P(t n )}

[0009] P min =min{P(t1),P(t2),P(t3),...,P(t i ),P(t i+1 ),...,P(t n )}

[0010] Wherein, icorr(ti)' is the normalized corrosion current density time series data at time ti, P(ti)' is the normalized polarization potential time series data at time ti, icorrmax is the maximum value of the corrosion current density time series data from time t1 to tn, icorrmin is the minimum value of the corrosion current density time series data from time t1 to tn, Pmax is the maximum value of the polarization potential time series data from time t1 to tn, and Pmin is the minimum value of the polarization potential time series data from time t1 to tn; the normalized corrosion current density time series data and the normalized polarization potential time series data from time t1 to tn can be obtained according to the normalization processing method.

[0011] Step 2: Set the initial parameters of the Long Short-Term Memory (LSTM) network. Based on the corrosion current density time series data prediction of the LTM network, use Tr={(icorr(t1)',P(t1)'),(icorr(t2)',P(t2)'),(icorr(t3)',P(t3)'),…,(icorr(ti)',P(ti)')} as the training set and Te={(icorr(t+1)',P(ti+1)'),…,(icorr(tn)',P(tn)')} as the test set to establish an LTM network prediction model. The time series length ratio of the training set to the test set is 7:3. Continuously update the LTM network structure according to the output value of the training process until the iteration conditions are met to complete the training process of the LTM network.

[0012] Step 3: Substitute the test samples {(icorr(t+1)',P(ti+1)'),…,(icorr(tn)',P(tn)')} into the trained Long Short-Term Memory network, and complete the establishment of the corrosion current density time series prediction by evaluating the output results;

[0013] Step 5: Calculate the correlation coefficient R between the polarization potential time series signal and the corrosion current density time series signal, the historical mean value of polarization potential μP, the historical mean value of corrosion current density μi, the historical standard deviation of polarization potential σP, and the historical standard deviation of corrosion current density σi.

[0014] Step 4: Obtain the polarization potential monitoring signal of the subway operation, input the polarization potential monitoring signal into the established long short-term memory network prediction model, obtain the predicted value P of the polarization potential signal, and obtain the calculation formula for the predicted value icorr of the corrosion current density signal:

[0015]

[0016] Further technical solution: The corrosion current density time series data is measured by three sets of three-electrode systems in the all-solid-state electrochemical corrosion sensor. The corrosion current densities at time ti are icorr1(ti), icorr2(ti), and icorr3(ti), respectively. The corrosion current density time series data at time ti is as follows:

[0017]

[0018] A further technical solution: The duration of the small cycle is determined by the corrosion weight loss Δm in the 0.1 mol / L electrolyte solution reaching 0.1 m, which is equivalent to the corrosion weight loss being 10% of the original weight m of the specimen. The corresponding corrosion time is the duration of the small cycle.

[0019] The corrosion weight loss at different corrosion times was measured, denoted as {Δm1,T1},{Δm2,T2},{Δm3,T3},……,{Δmn,Tn}, and the relationship between corrosion time and corrosion weight loss was fitted, namely:

[0020] Δm=f(T i )

[0021] Based on the fitting equation, 10% of the initial weight, which is 0.1m, is substituted into the fitting equation to obtain the time length of the small cycle.

[0022] Beneficial effects:

[0023] This invention provides a time-series prediction method for corrosion current density of subway steel fiber reinforced concrete structures by integrating polarization potential monitoring signals. It can effectively predict the development time-series law of stray current corrosion by analyzing the changes in stray current corrosion current density over a short period of time, providing effective support for early warning and timely protection against stray current corrosion of subway steel fiber reinforced concrete structures. Attached Figure Description

[0024] Figure 1 This is a schematic diagram of the process proposed in this invention;

[0025] Figure 2 This is a schematic diagram of the simulation experimental system in the method for determining the short-cycle time length proposed in this invention. Detailed Implementation

[0026] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0027] The following section will replace the Long Short-Term Memory network with LSTM. A time-series prediction method for corrosion current density of subway steel fiber reinforced concrete structures that integrates polarization potential monitoring signals includes the following steps:

[0028] Step 1: Measure the corrosion current density time series data {icorr(t1),icorr(t2),icorr(t3),…,icorr(ti),icorr(ti+1),…,icorr(tn)} and polarization potential time series data {P(t1),P(t2),P(t3),…,P(ti),P(ti+1),…,P(tn)} within the small period T1. n The corrosion was measured at t using a three-electrode system within the all-solid-state electrochemical corrosion sensor. iCorrosion current density i at time corr 1 (t i ), i corr 2 (t i ) and i corr 3 (t i All-solid-state electrochemical corrosion sensor t i The corrosion current density output value at time t is:

[0029]

[0030] And so on, to obtain t1 to t2. n Corrosion current density at time {i corr (t1),i corr (t2),i corr (t3),…,i corr (t i ),…,i corr (t n )}, where i corr (t1),i corr (t2),i corr (t3),i corr (t i+1 ),i corr (t n ) are respectively t1, t2, t3, t i+1 and t n The corrosion current density output value of the all-solid-state electrochemical corrosion sensor at time t1~t n The duration of subway operation is one week.

[0031] Step 3: Short-cycle corrosion current density time series data {i corr (t1),i corr (t2),i corr (t3),…,i corr (t i ),i corr (t i+1 ),…,i corr (t n The polarization potential timing data and the planned potential timing data {P(t1),P(t2),P(t3),…,P(t)} are used. i ),P(t i+1 ),…,P(t n The preprocessing of corrosion current density and polarization potential time series data is as follows:

[0032]

[0033] P max =max{P(t1),P(t2),P(t3),...,P(t i ),P(t i+1 ),...,P(t n )}

[0034] P min =min{P(t1),P(t2),P(t3),...,P(t i ),P(t i+1 ),...,P(t n )}

[0035] i corr (t i )' and P(t i )' are respectively t i The normalized output values ​​of corrosion current density and polarization potential of the all-solid-state electrochemical corrosion sensor, i corr max and i corr min They are respectively {i corr (t1),i corr (t2),i corr (t3),…,i corr (t i ),…,i corr (t n The maximum and minimum values ​​in}, P max and P min They are respectively {P(t1), P(t2), P(t3), ..., P(t i ),…,P(t n The maximum and minimum values ​​in} can be obtained. Similarly, this allows us to obtain t1 to t... n Time-series data of corrosion current density after time-normalization: {i corr (t1)',i corr (t2)',i corr (t3)',…,i corr (t i )',i corr (t i+1 )',…,i corr (t n The data includes the normalized polarization potential timing data: {P(t1)',P(t2)',P(t3)',…,P(t)'}, and the normalized polarization potential timing data. i )',P(t i+1 )',…,P(t n )'}, where i corr (t1)',i corr(t2)',i corr (t3)',i corr (t i+1 )',i corr (t n )' are respectively t1, t2, t3, t i+1 and t n The time-normalized corrosion current density output values ​​of the all-solid-state electrochemical corrosion sensor, where P(t1)', P(t2)', P(t3)', P(t i+1 )',P(t n )' are respectively t1, t2, t3, t i+1 and t n The polarization potential output value after time normalization.

[0036] Step 3: Set the initial parameters of the LSTM network.

[0037] Step 4: Time-series prediction of corrosion current density based on LSTM. (Using T...) r ={(i corr (t1)',P(t1)'),(i corr (t2)',P(t2)'),(i corr (t3)',P(t3)'),…,(i corr (t i )',P(t i )')} is the training set, and T is the training set. e ={(i corr (t +1 )',P(t i+1 )'),…,(i corr (t n )',P(t n Using the test set as the training set, an LSTM prediction model is built, where the time series length ratio of the training set to the test set is 7:3. The LSTM structure is continuously updated based on the output values ​​during the training process until the iteration conditions are met, completing the training process of the LSTM network.

[0038] Step 5: Transfer the test sample {(i corr (t +1 )',P(t i+1 )'),…,(i corr (t n )',P(t n Substitute the trained LSTM network into the output results and establish the corrosion current density time series prediction.

[0039] Step 6: Calculate the correlation coefficient R between the polarization potential time series signal and the corrosion current density time series signal, and the historical mean value μ of the polarization potential. PHistorical average corrosion current density μ i Polarization potential history standard deviation σ P Historical standard deviation of corrosion current density σ i .

[0040] Step 7: Given that polarization potential monitoring signals can be directly obtained at the subway site, the on-site polarization potential monitoring signals are input into the established prediction model to obtain the predicted value P of the polarization potential signal, and the predicted value icorr of the corrosion current density signal is obtained through the following method:

[0041]

[0042] Step 8: Based on the dangerous threshold of stray current corrosion current density of steel fiber reinforced concrete structures, if the predicted value is higher than the dangerous threshold, effective corrosion protection measures should be carried out in the corresponding area to alleviate and weaken the stray current corrosion effect.

[0043] According to claim 1, the method for predicting the corrosion current density of subway steel fiber reinforced concrete structures by integrating polarization potential monitoring signals is characterized in that the specific time length T1 of the small cycle in step 1 can be obtained by simulating the corrosion weight loss within 1 hour in a laboratory environment. Specifically, when the corrosion weight loss Δm reaches 0.1m in a 0.1mol / L electrolyte solution, the corrosion weight loss is 10% of the original weight m of the specimen. The corresponding corrosion time is the small cycle time length T1 that needs to be determined in step 1.

[0044] The corrosion weight loss at different corrosion times is calculated as {Δm1,T1},{Δm2,T2},{Δm3,T3},……,{Δm n The relationship between corrosion time and corrosion weight loss is fitted using the formula: ,Tn}.

[0045] Δm=f(T i )

[0046] Based on the fitting equation, 10% of the initial weight, which is 0.1m, is substituted into the fitting equation to obtain the time length T of the small cycle.

[0047] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A time-series prediction method for corrosion current density of subway steel fiber reinforced concrete structures by integrating polarization potential monitoring signals, characterized in that, Includes the following steps: Step 1: Set the subway's operating cycle duration to small cycles such as t1, t2, t3, ..., ti, ..., tn. Measure the corrosion current density time series data for the small cycles t1 to tn as icorr(t1), icorr(t2), icorr(t3), ..., icorr(ti), ..., icorr(tn). Measure the polarization potential time series data for the small cycles t1 to tn as P(t1), P(t2), P(t3), ..., P(ti), ..., P(tn). The calculation formula for the normalization processing method of the corrosion current density time series data and the polarization potential time series data is as follows: P max =max{P(t1),P(t2),P(t3),...,P(t i ),P(t i+1 ),...,P(t n )} P min =min{P(t1),P(t2),P(t3),...,P(t i ),P(t i+1 ),...,P(t n )} Wherein, icorr(ti)' is the normalized corrosion current density time series data at time ti, P(ti)' is the normalized polarization potential time series data at time ti, icorrmax is the maximum value of the corrosion current density time series data from time t1 to tn, icorrmin is the minimum value of the corrosion current density time series data from time t1 to tn, Pmax is the maximum value of the polarization potential time series data from time t1 to tn, and Pmin is the minimum value of the polarization potential time series data from time t1 to tn. According to the normalization method, the normalized corrosion current density time series data and the normalized polarization potential time series data at times t1 to tn can be obtained. Step 2: Set the initial parameters of the Long Short-Term Memory (LSTM) network. Based on the corrosion current density time series data prediction of the LTM network, use Tr={(icorr(t1)',P(t1)'),(icorr(t2)',P(t2)'),(icorr(t3)',P(t3)'),…,(icorr(ti)',P(ti)')} as the training set and Te={(icorr(t+1)',P(ti+1)'),…,(icorr(tn)',P(tn)')} as the test set to establish an LTM network prediction model. The time series length ratio of the training set to the test set is 7:

3. Continuously update the LTM network structure according to the output value of the training process until the iteration conditions are met to complete the training process of the LTM network. Step 3: Substitute the test samples {(icorr(t+1)',P(ti+1)'),…,(icorr(tn)',P(tn)')} into the trained Long Short-Term Memory network, and complete the establishment of the corrosion current density time series prediction by evaluating the output results; Step 5: Calculate the correlation coefficient R between the polarization potential time series signal and the corrosion current density time series signal, the historical mean value of polarization potential μP, the historical mean value of corrosion current density μi, the historical standard deviation of polarization potential σP, and the historical standard deviation of corrosion current density σi. Step 4: Obtain the polarization potential monitoring signal of the subway operation, input the polarization potential monitoring signal into the established long short-term memory network prediction model, obtain the predicted value P of the polarization potential signal, and obtain the calculation formula for the predicted value icorr of the corrosion current density signal:

2. The method for time-series prediction of corrosion current density in subway steel fiber reinforced concrete structures based on fused polarization potential monitoring signals, as described in claim 1, is characterized in that... The corrosion current density time series data were measured at time ti using three sets of three-electrode systems within the all-solid-state electrochemical corrosion sensor. The corrosion current densities at time ti were icorr1(ti), icorr2(ti), and icorr3(ti), respectively. The corrosion current density time series data at time ti were as follows:

3. The method for time-series prediction of corrosion current density in subway steel fiber reinforced concrete structures based on integrated polarization potential monitoring signals, as described in claim 1, is characterized in that... The duration of the small cycle is determined by the corrosion weight loss Δm in the 0.1 mol / L electrolyte solution reaching 0.1 m, which is equivalent to the corrosion weight loss being 10% of the original weight m of the specimen. The corresponding corrosion time is the duration of the small cycle. The corrosion weight loss at different corrosion times was measured, denoted as {Δm1,T1},{Δm2,T2},{Δm3,T3},……,{Δmn,Tn}, and the relationship between corrosion time and corrosion weight loss was fitted, namely: Δm=f(T i ) Based on the fitting equation, 10% of the initial weight, which is 0.1m, is substituted into the fitting equation to obtain the time length of the small cycle.

Images