A method for predicting the pressure of OMEGA waterstop strips at the joints of immersed tunnels

Abstract

The present invention discloses a method for predicting the OMEGA waterstop pressure of a joint of an immersed tube tunnel. First, monitoring data of a target immersed tube tunnel is acquired, and the data include temperature, humidity, temperature gradient, triaxial acceleration and waterstop pressure. The monitoring data are preprocessed and normalized to form a data set of a time series. Then, data features of the mean, median, mode, standard deviation and variance of each monitoring data are obtained, and frequency features of each frequency component of the monitoring data are obtained. Finally, an LSTM model and a Transformer model are constructed, and the temperature, humidity, temperature gradient, triaxial acceleration and waterstop pressure of a previous time period are used as input variables, and the waterstop pressure of a subsequent time period is used as a predicted output variable. The model is trained using the data set and the obtained data features and frequency features. After the training is completed, the prediction results of the two models are fused to generate a final prediction result.

Description

Technical Field

[0001] The invention belongs to the technical field of intelligent monitoring of immersed tube tunnel engineering structures, and in particular relates to a method for predicting the pressure of an OMEGA waterstop belt at a joint of an immersed tube tunnel. Background Art

[0002] OMEGA waterstop was first used in underground engineering and hydraulic structures in Europe. It is a waterproof material specially used for joints of engineering structures. It is usually used for joint treatment that requires waterproofing, such as submarine tunnels, underground engineering and hydraulic structures. In modern submarine tunnel construction, the pressure monitoring and prediction of OMEGA waterstop joints has always been a key issue in engineering safety management. However, the existing monitoring and prediction methods face many challenges in practical applications, especially when dealing with complex and changing construction environments. Traditional OMEGA waterstop pressure prediction systems mainly rely on single or limited environmental parameters, such as temperature and humidity. These systems usually use simple linear regression models or empirical formulas for prediction. However, the construction environment of submarine tunnels is extremely complex, involving the interaction of multiple environmental and dynamic factors. These factors do not exist independently, but affect the pressure state of the waterstop through complex nonlinear relationships. Due to the simplicity of the model structure, traditional methods are difficult to capture the interaction between these factors, resulting in low prediction accuracy. In addition, the existing monitoring and prediction systems also have lags in data acquisition and processing. Especially in the face of rapidly changing environmental conditions, the system often cannot adjust the sampling frequency or processing model in time, and thus cannot provide real-time accurate predictions. This lag increases safety risks during the construction process and may also result in the inability to take timely measures in emergency situations.

[0003] In addition to the lack of prediction accuracy and real-time performance, the existing monitoring system also lacks effective dynamic adjustment capabilities when dealing with drastic changes in the environment. For example, when sudden temperature changes occur during construction or when dynamic loads fluctuate significantly, the existing system usually relies on pre-set fixed parameters for prediction and cannot automatically adjust the model's parameter settings based on real-time monitoring data. This static model configuration often distorts the system's prediction results when facing extreme conditions and cannot reflect actual pressure changes.

[0004] In short, the current joint OMEGA waterstop pressure monitoring and prediction technology still has significant shortcomings in coping with complex and changeable construction environments and ensuring construction safety, and it is urgent to introduce more advanced technical means for improvement and optimization. Summary of the invention

[0005] The present invention mainly aims to overcome the deficiencies of the prior art and provides a method for predicting the pressure of an OMEGA waterstop at a joint of an immersed tube tunnel.

[0006] The present invention is achieved through the following technical solutions:

[0007] A method for predicting the pressure of an OMEGA waterstop at a joint of an immersed tube tunnel comprises the following steps:

[0008] Step 1, obtaining monitoring data of the target immersed tube tunnel, including: temperature, humidity, temperature gradient, triaxial acceleration and waterstop pressure;

[0009] Step 2: preprocess and normalize the monitoring data obtained in step 1 to form a time series data set;

[0010] Step 3, perform descriptive statistical analysis on the monitoring data to obtain the data characteristics of the mean, median, mode, standard deviation and variance of each monitoring data;

[0011] Step 4, segment each monitoring data in time series by moving the window;

[0012] Step 5, converting the data of each time period obtained after segmentation in step 4 into frequency domain data through short-time Fourier transform;

[0013] Step 6, for the frequency domain data of each time period obtained in step 5, the power spectrum density within its frequency range is calculated to obtain the frequency characteristics of each frequency component of the monitoring data;

[0014] Step 7, construct the LSTM model and the Transformer model, take the temperature, humidity, temperature gradient, triaxial acceleration, and waterstop pressure of the previous time period as input variables, and take the waterstop pressure of the next time period as the predicted output variable; use the data set of step 2, the data features obtained in step 3, and the frequency features obtained in step 6 to train the LSTM model and the Transformer model; after the model training is completed, fuse the prediction results of the LSTM model and the Transformer model through the weighted average method to generate the final prediction result.

[0015] In the above technical solution, in step 1, an adaptive sampling technique is used to dynamically adjust the sampling frequency of the data based on the change rate of the historical monitoring data, so as to determine the appropriate sampling intervals for each type of monitoring data; the faster the change rate of the historical monitoring data, the faster the corresponding sampling frequency is adjusted.

[0016] In the above technical solution, in step 2, the preprocessing includes missing value processing and outlier detection and correction to ensure data quality and integrity.

[0017] In the above technical solution, in step 2, the missing value processing includes: 1) if the missing rate of monitoring data at a certain time point exceeds the set threshold, the data at that time point will be deleted; 2) if continuous missing values ​​appear within the sampling period, choose to use the first valid value in front or behind to fill the missing data.

[0018] In the above technical solution, in step 2, outlier detection and correction: outlier detection is performed using a Z-score-based method, and the identified abnormal data is corrected.

[0019] In the above technical solution, in step 2, the normalization process is to scale all data to the interval [0, 1] to ensure that all data have the same scale.

[0020] In the above technical solution, in step 7, when the prediction results of the LSTM model and the Transformer model are fused by the weighted average method, the weights of the two models are allocated according to the inverse proportion of the errors of the two models, and the model with smaller error is allocated a higher weight.

[0021] In the above technical solution, a method combining grid search and Bayesian optimization is used to optimize the model parameters of the LSTM model and the Transformer model.

[0022] In the above technical solution, when the change rate of the monitored temperature or humidity or three-axis acceleration data is greater than the set threshold, the LSTM model and the Transformer model are retrained and the model parameters are adjusted to ensure that the model can adapt to the new environmental conditions in real time and maintain the accuracy of the prediction.

[0023] The advantages and beneficial effects of the present invention are:

[0024] The present invention provides a joint OMEGA waterstop pressure prediction method based on LSTM and Transformer models, which realizes accurate prediction and real-time monitoring by comprehensively analyzing the influence of key independent variables such as temperature, humidity, temperature gradient and triaxial acceleration on the waterstop pressure. Compared with traditional methods, the present invention has significant beneficial effects. First, by adopting an adaptive sampling strategy, the data acquisition frequency can be dynamically adjusted according to environmental changes, ensuring accurate data capture when key environmental conditions change drastically, improving the sensitivity of the system and reducing the risk of data omission. Secondly, by combining LSTM and Transformer models, long-term dependencies and complex nonlinear relationships in time series data are effectively captured, improving the accuracy and stability of prediction. The LSTM network is good at processing long-term dependencies, ensuring that the prediction accuracy is maintained during long-term monitoring, while the Transformer model accurately identifies the interactions between the respective variables through the self-attention mechanism, and further optimizes the model performance. In addition, the dynamic parameter adjustment mechanism of the present invention can monitor the changes of key variables in the construction process in real time, automatically adjust the model parameters, and ensure that the prediction is always accurate. The present invention is designed with a real-time warning function. When it is predicted that the waterstop pressure is close to or exceeds the safety threshold, the system immediately sends out a warning signal, which is transmitted in real time to the terminal device of the engineering management team through the Internet of Things technology. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 The present invention is a flow chart of the method for predicting the pressure of the OMEGA waterstop strip at the joint of the immersed tube tunnel.

[0026] Figure 2 A data graph showing the original data and predicted data of the OMEGA waterstop pressure of the immersed tube tunnel joint provided in an embodiment of the present invention.

[0027] For ordinary technicians in this field, other relevant drawings can be obtained based on the above drawings without any creative work. DETAILED DESCRIPTION

[0028] In order to enable those skilled in the art to better understand the technical solution of the present invention, the technical solution of the present invention is further described below in conjunction with specific embodiments.

[0029] A method for predicting the pressure of OMEGA waterstop strips at the joints of immersed tube tunnels, see Appendix Figure 1 , including the following steps.

[0030] Step 1: Deploy multiple high-precision sensor networks in the target submarine immersed tube tunnel project to monitor key environmental and structural data, including the following monitoring data.

[0031] Temperature (T): Calculated by temperature sensor, with a data range of 15-40°C. The temperature sensor is arranged on the top side wall of the carriageway of the immersed tube tunnel.

[0032] Humidity (H): Measured by humidity sensors, the humidity range is 50-100%. The humidity sensors are arranged on the top side wall of the carriageway of the immersed tube tunnel.

[0033] Temperature gradient ( ): It is calculated by differential calculation of temperature sensors arranged at different depths. The data range is -5°C to 5°C, recorded once every hour, with an accuracy of ±0.1°C / m.

[0034] Three-axis acceleration ( A x , A y , A z ): obtained through an acceleration sensor with a value of 0.001-1.5g. The acceleration sensor is arranged on the side wall of the pipe segment of the immersed tube tunnel.

[0035] Waterstop pressure (P): The collection range is -1-10 MPa, recorded once a minute by the pressure sensor with an accuracy of ±0.01 MPa.

[0036] These monitoring data are stored in a structured manner in the database. The system processes these data in a time series and associates each record with a specific timestamp to form a highly structured data set. , the data monitored by each sensor is , which the system stores as:

[0037] ;

[0038] Furthermore, for the optimization of sampling intervals, adaptive sampling technology is used to dynamically adjust the sampling frequency of data based on the change rate of historical monitoring data, so as to determine the appropriate sampling intervals for each type of monitoring data. The faster the change rate of historical monitoring data, the faster the corresponding sampling frequency is adjusted to enhance the sensitivity and real-time performance of data collection.

[0039] This process can be expressed by the following formula: ;

[0040] in, is the sampling frequency, is the sampling time interval, is the rate of change of the data.

[0041] Step 2: preprocess and normalize the monitoring data obtained in step 1 to form a time series data set.

[0042] Preprocessing, including missing value handling and outlier detection and correction, is performed to ensure data quality and integrity.

[0043] 1. Missing value processing: If the missing rate of monitoring data at a certain time point exceeds 30%, the data at that time point will be deleted to avoid significant impact on the overall data set. For a small number of consecutive missing data (such as missing 3-5 points), forward or backward filling methods are used to fill in. Specifically, if If the data at a certain moment is missing, you can use the previous valid data or filling:

[0044] .

[0045] 2. Outlier detection and correction: The Z-score-based method is used for outlier detection. The outlier judgment formula is:

[0046] ;

[0047] in is the sample mean, is the sample standard deviation; if , then the point is considered as an outlier and is corrected.

[0048] Example: Assume that the original temperature data is: T(t)=[22.5, 23.0, 22.8, missing, 23.2, outlier], the processed data is: T processed (t)=[22.5, 23.0, 22.8, 22.9 (filled), 23.2, 23.0 (corrected)]. Through these processing, the original data are converted into a high-quality data set suitable for the training and analysis of the waterstop pressure prediction model, ensuring the accuracy and reliability of the subsequent waterstop pressure prediction model prediction.

[0049] Normalization processing is to scale all data to the interval [0, 1] to ensure that all data have the same scale, thereby improving the performance and stability of the waterstop pressure prediction model.

[0050] Furthermore, the data set obtained in step 2 can be encrypted and stored in a cloud database, and encrypted and managed using blockchain technology to ensure that the data cannot be tampered with and is traceable.

[0051] In addition, the data storage adopts the NoSQL database structure, which can efficiently process and query complex time series data. For example, each data record can be represented as:

[0052] .

[0053] The system will regularly back up data to an offsite server to prevent data loss. The data processing status is updated in real time through the cloud console, and the status is marked as "processing", "processed" or "error".

[0054] Step 3: Perform descriptive statistical analysis on the monitoring data to obtain data characteristics such as the mean, median, mode, standard deviation and variance of each monitoring data, so as to quantify the central trend and dispersion characteristics of the data set, which can provide a training basis for the subsequent waterstop pressure prediction model.

[0055] 1. Quantifying the central tendency includes the following data characteristics.

[0056] 1. Calculate the mean: reflects the central tendency of the data;

[0057] ;

[0058] in, Indicates data, The total amount of data.

[0059] 2. Calculate the median: reflects the middle point of the data.

[0060] 3. Calculate the mode: the value that appears most frequently in the data.

[0061] 2. Quantify the degree of discreteness, including the following data characteristics.

[0062] 1. Calculate the standard deviation: reflects the volatility of the data;

[0063] ;

[0064] In the formula, is the mean.

[0065] 2. Calculate the variance: reflects the difference between the data and the mean. The calculation formula is:

[0066] .

[0067] Step 4: Segment the time series of each monitoring data through a moving window, so as to identify the changes of the main influencing factors in the data of different time periods in the subsequent steps.

[0068] By introducing moving window analysis, a fixed-length window is defined on the time series data, and the window is moved on the entire time series to locally analyze the characteristics of different time periods. The effect of different variables on the waterstop pressure can be identified within the dynamic window.

[0069] Assume that the time series data is , the window length is , the window position is , then the subsequence within the window is:

[0070] ;

[0071] The operation of moving the window can be expressed as:

[0072] ;

[0073] in, is the moving step size of the window.

[0074] Step 5: Convert the data of each time period obtained after segmentation in step 4 into frequency domain data through short-time Fourier transform (STFT).

[0075] Specifically, for a time series data , STFT performs the following mathematical expression:

[0076] ;

[0077] in, It's time and frequency The STFT result on , is a window function (such as a Hanning window or a Gaussian window) used to localize the time domain signal. is the complex exponential term of the Fourier transform.

[0078] Step 6: For the frequency domain data of each time period obtained in step 5, calculate its frequency The power spectral density (PSD) within the range is calculated to obtain the frequency characteristics of each frequency component of the monitoring data.

[0079] frequency The power spectral density (PSD) within the range is calculated as:

[0080] .

[0081] Step 7, construct the LSTM model and the Transformer model, take the temperature, humidity, temperature gradient, triaxial acceleration, and waterstop pressure of the previous time period as input variables, and take the waterstop pressure of the next time period as the predicted output variable; use the data set of step 2, the data features obtained in step 3, and the frequency features obtained in step 6 to train the LSTM model and the Transformer model; after the model training is completed, fuse the prediction results of the LSTM model and the Transformer model through the weighted average method to generate the final prediction result.

[0082] In this step, the LSTM network can well capture the long-term dependencies in the input data; the Transformer model can process multi-dimensional input features through the self-attention mechanism, combine the sequential information in the time series, and predict the future displacement.

[0083] After the model training is completed, the prediction results of the LSTM model and the Transformer model are combined, and the weighted average method is used to further optimize the final prediction output. First, the error performance of the LSTM model and the Transformer model on the validation set is evaluated. Common evaluation indicators include mean square error (MSE), root mean square error (RMSE) or mean absolute error (MAE). The evaluation results of the example are as follows:

[0084] The MSE of the LSTM model is:

[0085] ;

[0086] The MSE of the Transformer model is:

[0087] ;

[0088] The weights of the two models are assigned according to the inverse ratio of their errors, that is, the model with smaller error is assigned higher weight. Assume that the inverse ratio of MSE is used as the weight calculation method:

[0089] The weights of the LSTM model are:

[0090] ;

[0091] The weights of the Transformer model are:

[0092] ;

[0093] Substituting the values ​​in the above formula, we get:

[0094] ;

[0095] .

[0096] Finally, the prediction results based on the LSTM and Transformer models and As the predicted value of future data, the final weighted average prediction result y predicted for:

[0097] .

[0098] In this embodiment, the prediction results of the LSTM model and the Transformer model are combined, and the prediction results of the final hose pressure are shown in the attached Figure 2 . Figure 2 In the timeline, 8 / 19-8 / 20 refers to the time period from August 19th to August 20th.

[0099] Furthermore, in order to further optimize the model parameters, the present invention adopts a method combining grid search and Bayesian optimization to optimize the model parameters. Among them, grid search is used to initially screen parameter combinations, while Bayesian optimization further refines parameter selection to maximize the prediction accuracy of the model.

[0100] Grid search searches for the best parameters for the model by traversing every parameter combination in a predefined parameter grid. Assume that the parameter grid is , the goal is to find the parameters So that the loss function Minimum: .

[0101] The objective function is estimated by Bayesian optimization based on a probability model and is optimized step by step by selecting the parameter combination with the largest expected improvement. The distribution of the loss function is estimated using a Gaussian process. ,in It is historical data. Expected improvement The formula is:

[0102] ;

[0103] in, represents the expectation or mean of a random variable, represents the maximum loss function value observed so far, is the current optimal parameter combination, is the estimate of the loss function under the optimal parameter combination. Through the above dual optimization strategy, the prediction performance of the model can be significantly improved, ensuring that the model can maintain a high level of prediction accuracy under different construction conditions.

[0104] Furthermore, based on the above dual optimization strategy, in order to cope with the real-time changing monitoring environment, the present invention designs a dynamic parameter adjustment mechanism. When the independent variables such as temperature, humidity or triaxial acceleration are monitored to change dramatically (that is, when the rate of change is greater than the set threshold), the model is automatically retrained and the model parameters are adjusted to ensure that the model can adapt to the new environmental conditions in real time and maintain the accuracy of the prediction.

[0105] Specifically, assuming that the current model parameters are , when the maximum value of the rate of change of temperature, humidity and triaxial acceleration Greater than the set threshold , then retrain the model and update the model parameters:

[0106] ;

[0107] in, is the rate of change of temperature, is the rate of change of humidity, is the rate of change of the three-axis acceleration. When , retraining is triggered and the model parameters are updated .

[0108] Furthermore, the present invention is designed with a real-time warning function. When it is predicted that the water stop pressure approaches or exceeds the safety threshold, the system immediately issues a warning signal. This warning signal is transmitted to the terminal device of the engineering management team in real time through the Internet of Things technology.

[0109] The present invention is described above by way of example. It should be noted that, without departing from the core of the present invention, any simple deformation, modification or other equivalent replacement that can be made by those skilled in the art without inventive effort falls within the protection scope of the present invention.

Claims

1. A method for predicting the pressure of OMEGA waterstop strips at the joints of an immersed tunnel, characterized in that: The following steps are involved: Step 1, obtaining monitoring data of the target immersed tube tunnel, including: temperature, humidity, temperature gradient, triaxial acceleration and waterstop pressure; Step 2: preprocess and normalize the monitoring data obtained in step 1 to form a time series data set; Step 3, perform descriptive statistical analysis on the monitoring data to obtain the data characteristics of the mean, median, mode, standard deviation and variance of each monitoring data; Step 4, segment each monitoring data in time series by moving the window; Step 5, converting the data of each time period obtained after segmentation in step 4 into frequency domain data through short-time Fourier transform; Step 6, for the frequency domain data of each time period obtained in step 5, the power spectrum density within its frequency range is calculated to obtain the frequency characteristics of each frequency component of the monitoring data; Step 7, construct the LSTM model and the Transformer model, take the temperature, humidity, temperature gradient, triaxial acceleration, and waterstop pressure of the previous time period as input variables, and take the waterstop pressure of the next time period as the predicted output variable; use the data set of step 2, the data features obtained in step 3, and the frequency features obtained in step 6 to train the LSTM model and the Transformer model; after the model training is completed, fuse the prediction results of the LSTM model and the Transformer model through the weighted average method to generate the final prediction result.

2. The method for predicting the pressure of the OMEGA waterstop strip at the joint of the immersed tube tunnel according to claim 1 is characterized in that: In step 1, an adaptive sampling technique is used to dynamically adjust the sampling frequency of the data based on the change rate of the historical monitoring data. The faster the change rate of the historical monitoring data, the faster the corresponding sampling frequency is adjusted.

3. The method for predicting the pressure of the OMEGA waterstop strip at the joint of the immersed tube tunnel according to claim 1 is characterized in that: In step 2, the preprocessing includes missing value processing and outlier detection and correction to ensure data quality and integrity.

4. The method for predicting the pressure of the OMEGA waterstop strip at the joint of the immersed tube tunnel according to claim 3 is characterized by: In step 2, missing value processing includes: 1) if the missing rate of monitoring data at a certain time point exceeds the set threshold, the data at that time point will be deleted; 2) if there are consecutive missing values ​​within the sampling period, choose to use the first valid value before or after to fill the missing data.

5. The method for predicting the pressure of the OMEGA waterstop strip at the joint of the immersed tube tunnel according to claim 3 is characterized in that: In step 2, outlier detection and correction: a Z-score-based method is used to detect outliers and correct the identified abnormal data.

6. The method for predicting the pressure of OMEGA waterstop strips at the joints of an immersed tube tunnel according to claim 1, characterized in that: In step 2, the normalization process is to scale all data to the interval [0, 1].

7. The method for predicting the pressure of OMEGA waterstop strips at the joints of an immersed tube tunnel according to claim 1, characterized in that: In step 7, when the prediction results of the LSTM model and the Transformer model are fused by the weighted average method, the weights of the two models are assigned according to the inverse ratio of the errors of the two models. The model with smaller error is assigned higher weight.

8. The method for predicting the pressure of OMEGA waterstop strips at the joints of an immersed tube tunnel according to claim 1, characterized in that: A combination of grid search and Bayesian optimization is used to optimize the model parameters of the LSTM model and the Transformer model.

9. The method for predicting the pressure of the OMEGA waterstop strip at the joint of the immersed tube tunnel according to claim 8, characterized in that: When the change rate of the monitored temperature, humidity or triaxial acceleration data is greater than the set threshold, the LSTM model and Transformer model are retrained and the model parameters are adjusted.

Images