Prestress steel cylinder concrete pipe protection layer service life prediction method and system

Abstract

The invention discloses a method and system for predicting the service life of a prestressed concrete cylinder pipe protective layer, and the method comprises the steps: building a diffusion permeation model of a corrosive medium in concrete based on a Fick diffusion law and an Arrhenius equation, calculating the performance degradation data of a material through the diffusion permeation model, and outputting a material degradation index; the method comprises the following steps: associating elastic modulus degradation of a protective layer with a crack growth rate through an improved CDM continuous damage mechanics theory, introducing a coupling factor of a dynamic stress field and a corrosion medium concentration, establishing a stress damage coupling model, inputting the structural stress monitoring data and a material degradation index into the stress damage coupling model for updating, and determining the structural stress monitoring data and the material degradation index. And establishing a life prediction model based on an LSTM (Long Short-Term Memory) network, inputting the protection layer environment parameters, the material degradation index and the stress distribution into the life prediction model for prediction, and outputting the residual life of the protection layer. And the protection layer prediction accuracy and efficiency are improved.

Description

Technical Field

[0001] This invention relates to the field of building material performance prediction technology, and in particular to a method and system for predicting the life of the protective layer of a prestressed steel cylinder concrete pipe. Background Technology

[0002] Prestressed concrete cylinder pipes (PCCPs) are widely used in long-distance water conveyance and municipal water supply and drainage. The durability of their protective layer directly affects the corrosion resistance and service life of the pipe body. Current life prediction methods, based on static models, only consider single environmental factors or fixed load conditions, making it difficult to reflect the complex dynamic coupling effects in actual operation. Traditional methods typically rely on accelerated corrosion tests in laboratories or empirical formulas, which suffer from long prediction cycles, low efficiency, and poor adaptability, failing to meet the needs of PCCP full life-cycle management. Summary of the Invention

[0003] The purpose of this invention is to solve the above-mentioned problems by designing a method and system for predicting the life of the protective layer of prestressed steel cylinder concrete pipe.

[0004] To achieve the above objectives, the technical solution of the present invention further includes the following steps in the above-mentioned method for predicting the life of the protective layer of a prestressed steel cylinder concrete pipe:

[0005] Sensors are used to collect environmental parameters of the protective layer, data on material performance degradation, and data on structural stress monitoring.

[0006] A diffusion and permeation model of corrosive media in concrete is established based on Fick's diffusion law and Arrhenius equation. The material performance degradation data is calculated using the diffusion and permeation model, and the material degradation index is output.

[0007] By using the improved CDM continuous damage mechanics theory, the degradation of the elastic modulus of the protective layer is correlated with the crack propagation rate. A coupling factor between the dynamic stress field and the concentration of the corrosive medium is introduced to establish a stress-damage coupling model. The stress monitoring data of the structure and the material degradation index are input into the stress-damage coupling model for updating, and the stress distribution is output.

[0008] A lifetime prediction model is established based on an LSTM (Long Short-Term Memory) network. The environmental parameters of the protective layer, the material degradation index, and the stress distribution are input into the lifetime prediction model for prediction, and the remaining lifetime of the protective layer is output.

[0009] Furthermore, in the aforementioned method for predicting the lifespan of a prestressed steel cylinder concrete pipe protective layer, the collection of multi-dimensional data from the medical supply chain, including the acquisition of environmental parameters, material performance degradation data, and structural stress monitoring data of the protective layer via sensors, comprises:

[0010] The collected data is processed using multi-sensor data fusion technology. Consistency checks are performed on data from multiple sensors with the same parameter to remove obviously abnormal data, resulting in cleaned data. The cleaned data is then fused using a Kalman filter algorithm.

[0011] Furthermore, in the aforementioned method for predicting the life of the protective layer of a prestressed steel cylinder concrete pipe, the diffusion and penetration model of the corrosive medium in concrete is established based on Fick's diffusion law and Arrhenius equation. The material performance degradation data is calculated using this diffusion and penetration model, and a material degradation index is output, including:

[0012] A diffusion and penetration model of corrosive media in concrete was established based on Fick's diffusion law and Arrhenius equation.

[0013] The degradation data of the material properties are calculated using a diffusion-permeability model. The medium is attached to the concrete surface and enters the interior through the surface pores, and then gradually moves to the deeper layers under the action of concentration difference.

[0014] Furthermore, in the aforementioned method for predicting the life of the protective layer of a prestressed steel cylinder concrete pipe, the step of establishing a diffusion and penetration model of the corrosive medium in concrete based on Fick's diffusion law and Arrhenius equation, calculating the material performance degradation data using the diffusion and penetration model, and outputting the material degradation index, further includes:

[0015] Based on the principle of the Arrhenius equation, a temperature regulation mechanism for diffusion rate is added to the model. When the ambient temperature rises, the molecular motion speed of the medium increases, and the permeation rate increases accordingly. When the temperature decreases, the permeation rate slows down.

[0016] The degree of material performance degradation is converted into a dimensionless index between 0 and 1, where 0 indicates that the material performance has not degraded and 1 indicates that the protective function has been completely lost. The material degradation index is then output.

[0017] Furthermore, in the aforementioned method for predicting the life of the protective layer of a prestressed concrete cylinder pipe, the improved CDM continuous damage mechanics theory correlates the degradation of the elastic modulus of the protective layer with the crack propagation rate, introduces a coupling factor between the dynamic stress field and the concentration of the corrosive medium, and establishes a stress-damage coupling model, including:

[0018] Long-term monitoring was conducted on the bends of the pipeline to record the rate of change of the concentration of corrosive medium under different stress states. At the same time, the changes in stress distribution under different medium concentrations were measured. By statistically analyzing the measured data, the coupling factor reflecting the strength of the synergistic effect between the two was determined.

[0019] By introducing a coupling factor into the connection between the two basic models, the output of the stress damage model can affect the penetration path of the corrosive medium, while the output of the corrosion damage model can change the mechanical property parameters of the material.

[0020] Furthermore, in the aforementioned method for predicting the life of the protective layer of a prestressed steel cylinder concrete pipe, the step of inputting the structural stress monitoring data and material degradation index into the stress-damage coupling model for updating and outputting the stress distribution further includes:

[0021] The collected structural stress monitoring data is screened to remove outliers caused by sensor failures. Based on the structural characteristics of the pipeline, the stress data of the monitoring points are divided into regions according to spatial location. The average value of multiple monitoring points in each region is taken as the representative stress value of that region.

[0022] The material degradation index is spatially distributed and corresponds to the stress monitoring area. The stress data of the same area is matched with the material degradation index of that area to output the stress distribution.

[0023] Furthermore, in the aforementioned method for predicting the life of the protective layer of a prestressed steel cylinder concrete pipe, the step of establishing a life prediction model based on an LSTM (Long Short-Term Memory) network, inputting the environmental parameters of the protective layer, the material degradation index, and the stress distribution into the life prediction model for prediction, and outputting the remaining life of the protective layer, includes:

[0024] Input nodes are set according to the type and number of input parameters. Environmental parameters include temperature, humidity and corrosive medium concentration monitoring items. Material degradation index includes elastic modulus reduction rate and crack density index. Stress distribution involves stress values ​​in different parts of the pipeline. Each parameter corresponds to an independent input node.

[0025] The number of hidden layers and nodes needs to be adjusted according to the data scale. Two hidden layers are set up, and the number of nodes in each layer is determined according to the historical data sample size to extract key features from the data.

[0026] Furthermore, in a prestressed steel cylinder concrete pipe protective layer life prediction system, the prestressed steel cylinder concrete pipe protective layer life prediction system includes the following modules:

[0027] The data acquisition module is used to collect environmental parameters of the protective layer, material performance degradation data, and structural stress monitoring data through sensors.

[0028] The permeation model building module is used to establish a diffusion and permeation model of corrosive media in concrete based on Fick's diffusion law and Arrhenius equation, calculate the material performance degradation data using the diffusion and permeation model, and output the material degradation index.

[0029] The coupling model establishment module is used to correlate the degradation of the elastic modulus of the protective layer with the crack propagation rate through the improved CDM continuous damage mechanics theory, introduce the coupling factor between the dynamic stress field and the concentration of the corrosive medium, establish a stress-damage coupling model, input the stress monitoring data of the structure and the material degradation index into the stress-damage coupling model for updating, and output the stress distribution.

[0030] The real-time remaining lifetime module is used to establish a lifetime prediction model based on an LSTM (Long Short-Term Memory) network. The environmental parameters of the protective layer, the material degradation index, and the stress distribution are input into the lifetime prediction model for prediction, and the remaining lifetime of the protective layer is output.

[0031] Furthermore, in a prestressed steel cylinder concrete pipe protective layer life prediction system, the coupling model establishment module includes the following sub-modules:

[0032] The sub-module is used to filter the collected structural stress monitoring data, remove outliers caused by sensor failures, and divide the stress data of the monitoring points into regions according to spatial location based on the structural characteristics of the pipeline. The average value of multiple monitoring points in each region is taken as the representative stress value of that region.

[0033] The output submodule is used to match the spatial distribution of the material degradation index with the stress monitoring area, matching the stress data of the same area with the material degradation index of that area, and outputting the stress distribution.

[0034] Furthermore, in a prestressed steel cylinder concrete pipe protective layer life prediction system, the coupling model establishment module includes the following sub-modules:

[0035] The input submodule is used to set input nodes according to the type and number of input parameters. Environmental parameters include temperature, humidity and corrosive medium concentration monitoring items. Material degradation index includes elastic modulus reduction rate and crack density index. Stress distribution involves stress values ​​in different parts of the pipeline. Each parameter corresponds to an independent input node.

[0036] Set up sub-modules. The number of layers and nodes for the hidden layers needs to be adjusted according to the data scale. Set up 2 hidden layers, and the number of nodes in each layer is determined according to the historical data sample size. Extract key features from the data.

[0037] Its beneficial effects are as follows: 1. It strengthens the influence of long-term key factors and weakens the interference of short-term fluctuations, making the prediction results more consistent with the actual aging pattern, significantly improving the accuracy and efficiency of the remaining life prediction of the protective layer, and providing a reliable basis for engineering decisions. 2. It improves the utilization rate of data and avoids prediction bias caused by data quality issues. Attached Figure Description

[0038] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention.

[0039] Figure 1 This is a schematic diagram of the first embodiment of a method for predicting the life of the protective layer of a prestressed steel cylinder concrete pipe according to the present invention;

[0040] Figure 2 This is a schematic diagram of a second embodiment of a method for predicting the life of a protective layer for a prestressed steel cylinder concrete pipe according to an embodiment of the present invention;

[0041] Figure 3 This is a schematic diagram of the first embodiment of a prestressed steel cylinder concrete pipe protective layer life prediction system according to an embodiment of the present invention. Detailed Implementation

[0042] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0043] Those skilled in the art will understand that, unless specifically stated otherwise, the singular forms “a,” “an,” “the,” and “the” used herein may also include the plural forms. It should be further understood that the term “comprising” as used in this specification means the presence of features, integers, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof.

[0044] The present invention will now be described in detail with reference to the accompanying drawings. Figure 1 As shown, a method for predicting the life of the protective layer of a prestressed concrete cylinder pipe is disclosed, which includes the following steps:

[0045] Step 101: Collect environmental parameters of the protective layer, material property degradation data, and structural stress monitoring data through sensors;

[0046] Specifically, in this embodiment, multi-sensor data fusion technology is used to process the collected data, perform consistency checks on multiple sensor data with the same parameter, remove obviously abnormal data to obtain cleaned data, and then fuse the cleaned data using the Kalman filter algorithm.

[0047] Specifically,

[0048] (I) Sensor Selection

[0049] Environmental parameter acquisition sensor

[0050] Temperature and humidity sensor: Select a digital temperature and humidity sensor with an accuracy of ±0.5℃ (temperature) and ±2%RH (humidity), such as the SHT3x series, which can monitor the temperature and humidity changes of the environment around the pipeline in real time.

[0051] Chloride ion concentration sensor: Employs a chloride ion selective electrode sensor with a measurement range of 0-10000 mg / L and an accuracy of ±5%, used to monitor the concentration of corrosive media such as chloride ions in the environment.

[0052] Carbon dioxide concentration sensor: An infrared carbon dioxide sensor is selected, with a measurement range of 0-5000ppm and an accuracy of ±30ppm, to monitor the effect of carbon dioxide on concrete carbonation.

[0053] Material property degradation data acquisition sensor

[0054] Concrete resistivity sensor: A four-electrode concrete resistivity sensor is used, with a measurement range of 0-200kΩ·cm and an accuracy of ±5%. It reflects the degree of carbonation and corrosion of concrete by measuring changes in resistivity.

[0055] Ultrasonic sensor: An ultrasonic sensor with a frequency of 50kHz-2MHz is selected to detect defects and structural changes inside concrete and assess the damage to the material.

[0056] Rebound hammer: A digital rebound hammer is used, with a measurement range of 20-60MPa and an accuracy of ±1MPa. It is used to determine the strength of concrete surface and indirectly reflect the degradation of material properties.

[0057] Structural stress monitoring sensor

[0058] Strain gauges: Resistance strain gauges with a sensitivity coefficient of 2.0±1% and a measurement range of -2000-2000με are selected and attached to the key stress-bearing parts of the pipeline to monitor strain changes in the structure and then calculate stress.

[0059] Fiber Bragg grating sensor: Employs a fiber Bragg grating strain sensor with a measurement range of -1000-3000με and an accuracy of ±2με. It features advantages such as anti-electromagnetic interference and corrosion resistance, making it suitable for long-term stress monitoring.

[0060] (II) Sensor Arrangement

[0061] Sensor placement needs to take into account factors such as pipeline specifications, installation environment, and stress characteristics.

[0062] Environmental parameter sensors: A monitoring point is set up every 10 meters along the axial direction around the pipeline. Each monitoring point is equipped with a temperature and humidity sensor, a chloride ion concentration sensor and a carbon dioxide concentration sensor. The sensors should be kept away from heat sources, water sources and other sources of interference, and should avoid direct contact with the pipeline surface, about 30cm away from the pipeline surface.

[0063] Material property degradation sensor: On the surface of the pipe, a concrete resistivity sensor and an ultrasonic sensor are arranged every 1 meter along the circumference. The rebound hammer measuring points are evenly distributed on the surface of the pipe, with 3-5 measuring points per square meter.

[0064] Structural stress sensors: At critical stress points such as bends, joints, and diameter changes in pipelines, 3-5 strain gauges and 1-2 fiber Bragg grating sensors should be placed at each location. The strain gauges should be attached to the inner and outer surfaces of the pipeline, while the fiber Bragg grating sensors can be embedded in the concrete or attached to the surface.

[0065] (III) Data Validation and Correction

[0066] Multi-sensor data fusion technology is employed to process the collected data. For data from multiple sensors sharing the same parameter, a consistency check is first performed to remove obviously abnormal data. Then, a weighted average method or Kalman filtering algorithm is used to fuse the valid data, improving the accuracy and reliability of the data. Simultaneously, the sensors are regularly calibrated and maintained to ensure their measurement accuracy.

[0067] Step 102: Based on Fick's diffusion law and Arrhenius equation, establish a diffusion and penetration model of corrosive media in concrete, use the diffusion and penetration model to calculate material performance degradation data, and output the material degradation index;

[0068] Specifically, in this embodiment, a diffusion and penetration model of corrosive media in concrete is established based on Fick's diffusion law and Arrhenius equation;

[0069] The diffusion-permeability model is used to calculate the material performance degradation data. The medium is attached to the concrete surface and enters the interior through the surface pores. Then, under the influence of the concentration difference, it gradually moves to the deeper layers.

[0070] Based on the principle of the Arrhenius equation, a temperature regulation mechanism for diffusion rate is added to the model. When the ambient temperature rises, the molecular motion speed of the medium increases, and the permeation rate increases accordingly. When the temperature decreases, the permeation rate slows down.

[0071] The degree of material performance degradation is converted into a dimensionless index between 0 and 1, where 0 indicates that the material performance has not degraded and 1 indicates that the protective function has been completely lost. The material degradation index is then output.

[0072] Specifically,

[0073] (I) Basic Principles

[0074] The diffusion-permeability model is based on Fick's diffusion law and the Arrhenius equation. Fick's second law describes the diffusion of corrosive media in concrete, and its expression is:

[0075]

[0076] Where C is the concentration of the corrosive medium (mg / m³) 3 ), t is time (s), x is diffusion distance (m), and D is diffusion coefficient (m). 2 / s).

[0077] The Arrhenius equation reflects the effect of temperature on the diffusion coefficient, and its expression is:

[0078] D = D0exp(-Q / (RT))

[0079] Where D0 is the initial diffusion coefficient (m 2 / s), Q is the diffusion activation energy (J / mol), R is the gas constant (8.314 J / (mol·K)), and T is the absolute temperature (K).

[0080] (II) Model Establishment

[0081] Substituting the Arrhenius equation into Fick's second law, we obtain the diffusion equation for corrosive media that considers the effect of temperature:

[0082]

[0083] This equation allows for the calculation of corrosive medium concentrations at different times and locations. Based on the relationship between corrosive medium concentration and material performance degradation, material performance degradation data can be calculated, and a material degradation index can be output. The material degradation index can be represented by indicators such as concrete strength loss rate and elastic modulus reduction rate. Its calculation formula is determined based on specific material performance parameters. For example, using the elastic modulus reduction rate as the material degradation index I, then I = (E0 - E) / E0, where E0 is the initial elastic modulus and E is the current elastic modulus.

[0084] Step 103: By using the improved CDM continuous damage mechanics theory, the degradation of the elastic modulus of the protective layer is correlated with the crack propagation rate. A coupling factor between the dynamic stress field and the concentration of the corrosive medium is introduced to establish a stress-damage coupling model. The structural stress monitoring data and material degradation index are input into the stress-damage coupling model for updating, and the stress distribution is output.

[0085] Specifically, in this embodiment, the elbow of the pipeline is selected for long-term monitoring. The rate of change of the concentration of corrosive medium under different stress states is recorded. At the same time, the change of stress distribution under different medium concentrations is measured. By statistically analyzing the measured data, the coupling factor reflecting the strength of the synergistic effect between the two is determined.

[0086] By introducing a coupling factor into the connection between the two basic models, the output of the stress damage model can affect the penetration path of the corrosive medium, while the output of the corrosion damage model can change the mechanical property parameters of the material.

[0087] The collected structural stress monitoring data is screened to remove outliers caused by sensor failures. Based on the structural characteristics of the pipeline, the stress data of the monitoring points are divided into regions according to spatial location. The average value of multiple monitoring points in each region is taken as the representative stress value of that region.

[0088] The material degradation index is spatially distributed and corresponds to the stress monitoring area. The stress data of the same area is matched with the material degradation index of that area to output the stress distribution.

[0089] Specifically,

[0090] (I) Improved CDM Continuous Damage Mechanics Theory

[0091] The CDM (Continuous Damage Mechanics) theory describes the degree of material damage through a damage variable, D, which ranges from 0 (no damage) to 1 (complete damage). The improved CDM theory correlates the degradation of the elastic modulus of the protective layer with the crack propagation rate, arguing that the degradation of the elastic modulus is caused by the initiation and propagation of cracks within the material. This relationship can be expressed as:

[0092] E=E0(1-D)n

[0093] Where n is a material constant, determined experimentally.

[0094] The crack propagation rate da / dt has a power function relationship with the stress intensity factor amplitude ΔK, that is:

[0095] da / dt=C(ΔK)□

[0096] Where C and m are material constants, and ΔK is the stress intensity factor amplitude.

[0097] (II) Introduction of Coupling Factor

[0098] A coupling factor k is introduced to represent the dynamic stress field and the concentration of the corrosive medium. This coupling factor reflects the combined effect of the stress field and the concentration of the corrosive medium on material damage. The expression for the coupling factor k is:

[0099] k=k1σ+k2C

[0100] Where k1 and k2 are coefficients determined by fitting experimental data, σ is the stress (MPa), and C is the concentration of the corrosive medium (mg / m³). 3 ).

[0101] (III) Establishment of Stress-Damage Coupling Model

[0102] By introducing a coupling factor into the damage evolution equation, a stress-damage coupling model is established:

[0103] dD / dt=f(D,σ,C)=k·g(D)

[0104] Where g(D) is the damage evolution function, which is determined based on the material properties.

[0105] Structural stress monitoring data and material degradation index are input into the stress-damage coupling model, and the model is updated through iterative calculations to output the stress distribution.

[0106] Specifically,

[0107] (I) Actual Manifestations of Coupling Effect

[0108] The dynamic stress field and the concentration of corrosive media have a mutually reinforcing effect: concrete in stress concentration areas is more prone to cracking, and these cracks become channels for the rapid penetration of corrosive media, accelerating the diffusion of the media inside the protective layer; while high concentrations of corrosive media weaken the strength of concrete, making the material more susceptible to damage under the same stress, thereby changing the stress distribution.

[0109] (II) Methods for obtaining coupling factors

[0110] Long-term monitoring was conducted on key sections of typical pipelines (elbows and joints) to record the rate of change of corrosive medium concentration under different stress states (working stress during pipeline operation and changes in external soil pressure). Simultaneously, changes in stress distribution under different medium concentrations were measured. Statistical analysis of these measured data determined a coupling factor that reflects the strength of the synergistic effect between the two. For example, in areas of high stress and high medium concentration, a larger coupling factor was chosen to reflect a stronger damage effect.

[0111] III. Construction Process of Stress-Damage Coupling Model

[0112] (I) Basic Model Framework Construction

[0113] First, construct stress damage model and corrosion damage model respectively. The stress damage model takes the stress state of the material as input and outputs the degree of damage caused by stress; the corrosion damage model takes the medium concentration distribution as input and outputs the degree of material deterioration caused by corrosion.

[0114] (II) Integration of Coupling Mechanisms

[0115] By introducing a coupling factor into the connection link between the two basic models, the output of the stress damage model can affect the penetration path of the corrosive medium (stress-induced cracks will change the direction of medium diffusion), while the output of the corrosion damage model will change the mechanical property parameters of the material (the decrease in concrete strength after medium erosion will affect stress transmission), forming a closed-loop model with mutual feedback.

[0116] IV. Processing of Model Input Data

[0117] (I) Structural stress monitoring data processing

[0118] The collected structural stress monitoring data were screened to remove outliers caused by sensor malfunctions. Based on the structural characteristics of the pipeline, the stress data of the monitoring points were divided into regions according to spatial location (pipe top, pipe side, pipe bottom). The average value of multiple monitoring points in each region was taken as the representative stress value of that region to ensure that the data can accurately reflect the stress state of different parts of the pipeline.

[0119] (II) Adaptation of Material Degradation Index

[0120] The material degradation index output in step 2 is spatially distributed and matched with the stress monitoring area. That is, the stress data of the same area is matched with the material degradation index of that area, so that the model can simultaneously consider the material performance state and stress conditions of that area.

[0121] V. Model Update and Stress Distribution Output

[0122] (I) Model Dynamic Update Mechanism

[0123] Regularly (quarterly), newly collected stress monitoring data and material degradation indices are input into the model. By comparing the model's predicted results with the actual monitored damage to the protective layer (crack propagation level), the correlation parameters in the model are adjusted. For example, when the actual crack propagation rate in a certain area is faster than the model prediction, the weight of the coupling factor in that area is increased to make the model more closely match the actual damage process.

[0124] (II) Output Form of Stress Distribution

[0125] The stress distribution output by the model is presented in intuitive visualization charts, such as stress cloud maps of pipe cross-sections, with different colors used to mark areas of high and low stress: red indicates high stress areas and blue indicates low stress areas. It also includes explanations of the stress value range and variation trend for each key region, providing a clear basis for subsequent life prediction based on the stress state.

[0126] Step 104: Establish a lifetime prediction model based on LSTM long short-term memory network. Input the environmental parameters of the protective layer, the material degradation index and the stress distribution into the lifetime prediction model for prediction, and output the remaining lifetime of the protective layer.

[0127] Specifically, in this embodiment, input nodes are set according to the type and number of input parameters. Environmental parameters include temperature, humidity and corrosive medium concentration monitoring items. Material degradation index includes elastic modulus reduction rate and crack density index. Stress distribution involves stress values ​​in different parts of the pipeline. Each parameter corresponds to an independent input node.

[0128] The number of hidden layers and nodes needs to be adjusted according to the data scale. Two hidden layers are set up, and the number of nodes in each layer is determined according to the historical data sample size to extract key features from the data.

[0129] Specifically,

[0130] I. LSTM Network Selection Criteria

[0131] LSTM (Long Short-Term Memory) networks are particularly well-suited for processing data with time-series characteristics. The aging process of the protective layer of prestressed concrete cylinder pipes is precisely dynamic over time—environmental parameters (temperature, humidity, corrosive medium concentration) fluctuate with seasons and climate, while material degradation indices (strength reduction, elastic modulus decay) and stress distribution gradually evolve over time. This temporal correlation allows LSTM to effectively capture the dependencies between data from different stages, such as the accelerated corrosion of the protective layer in hot and humid summer conditions, and its lag correlation with subsequent material performance degradation, thus enabling more accurate prediction of remaining lifespan.

[0132] II. Core Steps in Model Building

[0133] (I) Network Structure Design

[0134] Input layer configuration: Input nodes are set according to the type and number of input parameters. Environmental parameters include monitoring items such as temperature, humidity, and concentration of corrosive media; material degradation index covers indicators such as elastic modulus reduction rate and crack density; stress distribution involves stress values ​​at different parts of the pipeline (top, side, and bottom). Each parameter corresponds to an independent input node to ensure that all types of data can be completely received by the network.

[0135] Hidden layer settings: The number of hidden layers and nodes needs to be adjusted according to the data scale. Initially, 2-3 hidden layers can be set, and the number of nodes in each layer is determined according to the historical data sample size (the larger the sample size, the more nodes can be added). The role of the hidden layers is to gradually extract key features from the data, such as identifying the key temperature and humidity ranges that accelerate corrosion from environmental parameters, and capturing the characteristics of high-stress areas prone to damage from stress distribution.

[0136] Output layer design: The output layer has only one node, which directly corresponds to the remaining lifespan of the protection layer. The unit is uniformly "years", which facilitates intuitive understanding and decision-making in engineering applications.

[0137] (II) Adaptation of Network Memory Mechanism

[0138] The core advantage of LSTM networks lies in their unique memory units, which can autonomously distinguish between important and secondary information. During model construction, the memory mechanism needs to be optimized to address the characteristics of protective layer aging: for long-term factors (continuous high-concentration corrosive media), the network strengthens the memory of their cumulative impact on material degradation; for short-term fluctuating factors (sudden increases in humidity due to short-term heavy rainfall), it weakens their interference with long-term lifetime prediction, avoiding the influence of accidental data fluctuations on prediction results.

[0139] III. Input Data Processing Methods

[0140] (I) Data Standardization

[0141] The collected environmental parameters, material degradation index, and stress distribution data use different units of measurement (temperature in degrees Celsius, stress in megapascals). These need to be standardized to convert them into a uniform numerical range (typically between 0 and 1). This prevents the network from overemphasizing any particular data type due to its excessively large magnitude (the concentration of corrosive media may be much higher than the temperature value), ensuring that all parameters play a balanced role in the model.

[0142] (II) Time Series Processing

[0143] All data are arranged chronologically to form continuous sequence segments. For example, using "months" as the time interval, the average environmental parameters, material degradation index, and stress distribution monitoring values ​​for each month are combined into a data set for a specific time node. Then, data sets from multiple consecutive months are concatenated to form a complete time series. The sequence length is determined based on the amount of historical data accumulated, generally requiring at least three years of continuous data to ensure the network can learn complete seasonal variations and aging trends.

[0144] (III) Abnormal Data Handling

[0145] Abnormal data (jump values ​​caused by sensor malfunctions, unreasonable values ​​caused by detection errors) are identified through manual verification and statistical analysis. For a small number of abnormal data, normal data from adjacent time points are used for smoothing correction; for consecutive abnormal segments, they need to be removed or supplemented based on the actual situation on site (temporary data interruptions during pipeline maintenance) to ensure that the data sequence input to the network is continuous and reliable.

[0146] IV. Life Prediction Implementation Process

[0147] (I) Real-time data input

[0148] Input the latest environmental parameters collected in Step 1 (temperature and humidity for the current quarter, concentration of corrosive media), the latest material degradation index output in Step 2 (recently measured rate of decrease in elastic modulus), and the latest stress distribution output in Step 3 (current stress values ​​at various parts of the pipeline) into the trained LSTM model in time series format. Ensure that the timestamps of the data correspond accurately during input to avoid prediction errors due to time misalignment.

[0149] (II) Model Reasoning Process

[0150] After receiving data, the network extracts features from the hidden layers and learns from historical data using a memory mechanism to automatically analyze the correlation between current data and past aging patterns. For example, when the concentration of the input corrosive medium suddenly increases, the network recalls historical data showing rapid material degradation after similar concentration changes, and calculates the impact of this change on the remaining lifespan based on the current stress distribution. The inference process requires no manual intervention and is completed autonomously by the model, typically taking only a few minutes.

[0151] (III) Dynamic Prediction and Update

[0152] Because the operating environment and condition of pipelines are constantly changing, the above prediction process needs to be repeated periodically (it is recommended to do so every six months) to update the prediction results with the latest data. If a prediction finds that the remaining lifespan is significantly shortened compared to the previous one (the shortening exceeds 20%), it is necessary to combine on-site inspections to investigate whether new damage factors have appeared (new cracks appear in the protective layer, or the surrounding environment suddenly deteriorates), and adjust the model input parameters according to the investigation results to ensure that the prediction results can reflect the actual condition of the pipeline in real time.

[0153] Specifically, by integrating multiphysics modeling and artificial intelligence algorithms, a comprehensive intelligent optimization system was constructed, encompassing material degradation prediction and equipment lifespan management, significantly improving the processing efficiency of semiconductor components. In corrosion control, etching parameters were dynamically adjusted based on the Fick-Arrhenius model, resulting in a 6 percentage point increase in wafer yield (e.g., a 12% reduction in etching cycle at one factory). Regarding equipment stability, combining CDM damage mechanics and stress monitoring precisely extended the CVD chamber maintenance cycle by 3 months, improving equipment OEE by 8%. Furthermore, optimizing the packaging protection layer maintenance strategy using an LSTM lifespan prediction model reduced maintenance costs by 20%. Experimental results show that the model prediction error is less than 5%, and the accuracy reaches 93% after production line data iteration, forming a closed loop of "data acquisition - model update - process optimization," providing key technical support for the semiconductor manufacturing industry's transition to predictive maintenance and adaptive processes.

[0154] Its beneficial effects are as follows: 1. It strengthens the influence of long-term key factors and weakens the interference of short-term fluctuations, making the prediction results more consistent with the actual aging pattern, significantly improving the accuracy and efficiency of the remaining life prediction of the protective layer, and providing a reliable basis for engineering decisions. 2. It improves the utilization rate of data and avoids prediction bias caused by data quality issues.

[0155] Please see Figure 2 In a method for predicting the life of the protective layer of a prestressed concrete cylinder pipe, the degradation of the elastic modulus of the protective layer is correlated with the crack propagation rate through an improved CDM continuous damage mechanics theory. A coupling factor between the dynamic stress field and the concentration of the corrosive medium is introduced to establish a stress-damage coupling model, which includes the following steps:

[0156] Step 201: Select the bends of the pipeline for long-term monitoring, record the rate of change of the concentration of corrosive medium under different stress states, and measure the change of stress distribution under different medium concentrations. By statistically analyzing the measured data, determine the coupling factor that reflects the strength of the synergistic effect between the two.

[0157] Step 202: Introduce a coupling factor into the connection link between the two basic models so that the output of the stress damage model can affect the penetration path of the corrosive medium, while the output of the corrosion damage model will change the mechanical property parameters of the material.

[0158] The above describes an embodiment of the method for predicting the life of the protective layer of a prestressed steel cylinder concrete pipe according to the present invention. Please refer to [link / reference]. Figure 3 In a prestressed steel cylinder concrete pipe protective layer life prediction system, the system includes the following modules:

[0159] The data acquisition module is used to collect environmental parameters of the protective layer, material performance degradation data, and structural stress monitoring data through sensors.

[0160] The permeation model building module is used to build a diffusion and permeation model of corrosive media in concrete based on Fick's diffusion law and Arrhenius equation. The diffusion and permeation model is used to calculate material performance degradation data and output the material degradation index.

[0161] The coupling model building module is used to correlate the degradation of the elastic modulus of the protective layer with the crack propagation rate through the improved CDM continuous damage mechanics theory. It introduces a coupling factor between the dynamic stress field and the concentration of the corrosive medium to establish a stress-damage coupling model. The structural stress monitoring data and material degradation index are input into the stress-damage coupling model for updating, and the stress distribution is output.

[0162] The real-time remaining lifetime module is used to build a lifetime prediction model based on the LSTM long short-term memory network. It inputs the environmental parameters of the protective layer, the material degradation index and the stress distribution into the lifetime prediction model for prediction, and outputs the remaining lifetime of the protective layer.

[0163] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely preferred examples and are not intended to limit the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for predicting the service life of the protective layer of a prestressed steel cylinder concrete pipe, characterized in that, The method for predicting the lifespan of the protective layer of the prestressed steel cylinder concrete pipe includes the following steps: Sensors are used to collect environmental parameters of the protective layer, data on material performance degradation, and data on structural stress monitoring. A diffusion and permeation model of corrosive media in concrete is established based on Fick's diffusion law and Arrhenius equation. The material performance degradation data is calculated using the diffusion and permeation model, and the material degradation index is output. By using the improved CDM continuous damage mechanics theory, the degradation of the elastic modulus of the protective layer is correlated with the crack propagation rate. A coupling factor between the dynamic stress field and the concentration of the corrosive medium is introduced to establish a stress-damage coupling model. The stress monitoring data of the structure and the material degradation index are input into the stress-damage coupling model for updating, and the stress distribution is output. A lifetime prediction model is established based on an LSTM (Long Short-Term Memory) network. The environmental parameters of the protective layer, the material degradation index, and the stress distribution are input into the lifetime prediction model for prediction, and the remaining lifetime of the protective layer is output.

2. The method for predicting the service life of the protective layer of a prestressed steel cylinder concrete pipe as described in claim 1, characterized in that, The process of collecting environmental parameters of the protective layer, material property degradation data, and structural stress monitoring data through sensors includes: The collected data is processed using multi-sensor data fusion technology. Consistency checks are performed on data from multiple sensors with the same parameter to remove obviously abnormal data, resulting in cleaned data. The cleaned data is then fused using a Kalman filter algorithm.

3. The method for predicting the service life of the protective layer of a prestressed steel cylinder concrete pipe as described in claim 1, characterized in that, The diffusion and permeation model of corrosive media in concrete, based on Fick's diffusion law and Arrhenius equation, is used to calculate the material performance degradation data and output the material degradation index, including: A diffusion and penetration model of corrosive media in concrete was established based on Fick's diffusion law and Arrhenius equation. The degradation data of the material properties are calculated using a diffusion-permeability model. The medium is attached to the concrete surface and enters the interior through the surface pores, and then gradually moves to the deeper layers under the action of concentration difference.

4. The method for predicting the service life of the protective layer of a prestressed steel cylinder concrete pipe as described in claim 1, characterized in that, The method for establishing a diffusion and permeation model of corrosive media in concrete based on Fick's diffusion law and Arrhenius equation, calculating material performance degradation data using the diffusion and permeation model, and outputting a material degradation index also includes: Based on the principle of the Arrhenius equation, a temperature regulation mechanism for diffusion rate is added to the model. When the ambient temperature rises, the molecular motion speed of the medium increases, and the permeation rate increases accordingly. When the temperature decreases, the permeation rate slows down. The degree of material performance degradation is converted into a dimensionless index between 0 and 1, where 0 indicates that the material performance has not degraded and 1 indicates that the protective function has been completely lost. The material degradation index is then output.

5. The method for predicting the service life of the protective layer of a prestressed steel cylinder concrete pipe as described in claim 1, characterized in that, The improved CDM continuous damage mechanics theory correlates the degradation of the elastic modulus of the protective layer with the crack propagation rate, introduces a coupling factor between the dynamic stress field and the concentration of the corrosive medium, and establishes a stress-damage coupling model, including: Long-term monitoring was conducted on the bends of the pipeline to record the rate of change of the concentration of corrosive medium under different stress states. At the same time, the changes in stress distribution under different medium concentrations were measured. By statistically analyzing the measured data, the coupling factor reflecting the strength of the synergistic effect between the two was determined. By introducing a coupling factor into the connection between the two basic models, the output of the stress damage model can affect the penetration path of the corrosive medium, while the output of the corrosion damage model can change the mechanical property parameters of the material.

6. The method for predicting the service life of the protective layer of a prestressed steel cylinder concrete pipe as described in claim 5, characterized in that, The step of inputting the structural stress monitoring data and material degradation index into the stress-damage coupling model for updating and outputting the stress distribution also includes: The collected structural stress monitoring data is screened to remove outliers caused by sensor failures. Based on the structural characteristics of the pipeline, the stress data of the monitoring points are divided into regions according to spatial location. The average value of multiple monitoring points in each region is taken as the representative stress value of that region. The material degradation index is spatially distributed and corresponds to the stress monitoring area. The stress data of the same area is matched with the material degradation index of that area to output the stress distribution.

7. The method for predicting the service life of the protective layer of a prestressed steel cylinder concrete pipe as described in claim 1, characterized in that, The lifetime prediction model based on the LSTM long short-term memory network is established by inputting the environmental parameters of the protective layer, the material degradation index, and the stress distribution into the lifetime prediction model for prediction, and outputting the remaining lifetime of the protective layer, including: Input nodes are set according to the type and number of input parameters. Environmental parameters include temperature, humidity and corrosive medium concentration monitoring items. Material degradation index includes elastic modulus reduction rate and crack density index. Stress distribution involves stress values ​​in different parts of the pipeline. Each parameter corresponds to an independent input node. The number of hidden layers and nodes needs to be adjusted according to the data scale. Two hidden layers are set up, and the number of nodes in each layer is determined according to the historical data sample size to extract key features from the data.

8. A system for predicting the lifespan of the protective layer of a prestressed steel cylinder concrete pipe, characterized in that, The prestressed steel cylinder concrete pipe protective layer life prediction system includes the following steps: The data acquisition module is used to collect environmental parameters of the protective layer, material performance degradation data, and structural stress monitoring data through sensors. The permeation model building module is used to establish a diffusion and permeation model of corrosive media in concrete based on Fick's diffusion law and Arrhenius equation, calculate the material performance degradation data using the diffusion and permeation model, and output the material degradation index. The coupling model establishment module is used to correlate the degradation of the elastic modulus of the protective layer with the crack propagation rate through the improved CDM continuous damage mechanics theory, introduce the coupling factor between the dynamic stress field and the concentration of the corrosive medium, establish a stress-damage coupling model, input the stress monitoring data of the structure and the material degradation index into the stress-damage coupling model for updating, and output the stress distribution. The real-time remaining lifetime module is used to establish a lifetime prediction model based on an LSTM (Long Short-Term Memory) network. The environmental parameters of the protective layer, the material degradation index, and the stress distribution are input into the lifetime prediction model for prediction, and the remaining lifetime of the protective layer is output.

9. The life prediction system for the protective layer of a prestressed steel cylinder concrete pipe as described in claim 8, characterized in that, The coupling model establishment module includes the following sub-modules: The sub-module is used to filter the collected structural stress monitoring data, remove outliers caused by sensor failures, and divide the stress data of the monitoring points into regions according to spatial location based on the structural characteristics of the pipeline. The average value of multiple monitoring points in each region is taken as the representative stress value of that region. The output submodule is used to match the spatial distribution of the material degradation index with the stress monitoring area, matching the stress data of the same area with the material degradation index of that area, and outputting the stress distribution.

10. The life prediction system for the protective layer of a prestressed steel cylinder concrete pipe as described in claim 8, characterized in that, The coupling model establishment module includes the following sub-modules: The input submodule is used to set input nodes according to the type and number of input parameters. Environmental parameters include temperature, humidity and corrosive medium concentration monitoring items. Material degradation index includes elastic modulus reduction rate and crack density index. Stress distribution involves stress values ​​in different parts of the pipeline. Each parameter corresponds to an independent input node. Set up sub-modules. The number of layers and nodes for the hidden layers needs to be adjusted according to the data scale. Set up 2 hidden layers, and the number of nodes in each layer is determined according to the historical data sample size. Extract key features from the data.

Images