TBM tunneling parameter abnormal data identification method and system
By identifying anomalous TBM tunneling parameters through polynomial chaotic expansion and data clustering, the problem of difficulty in identification by traditional methods is solved, enabling more accurate data modeling and analysis, adapting to low sampling rate characteristics, and providing reasonable data partitioning results.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- CHINA RAILWAY SHISIJU GROUP CORP
- Filing Date
- 2025-02-21
- Publication Date
- 2026-07-07
AI Technical Summary
The presence of anomalous data in the TBM tunneling parameter data leads to deviations in subsequent data modeling, analysis, and mining results. Traditional methods struggle to effectively identify these anomalous data, especially in highly interference environments.
Polynomial chaotic expansion (PCE) is used to characterize the nonlinear correlation between tunneling parameters. Data clustering is used to assign membership functions to the tunneling parameters. The Lagrange multiplier method is used to optimize and solve the data clustering objective function to identify abnormal data.
It effectively identifies abnormal data in TBM tunneling parameters, improves the accuracy of data modeling and analysis, avoids the "curse of dimensionality" problem, and provides better nonlinear characterization and interpretability.
Smart Images
Figure CN120105299B_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of data processing technology applied to TBM tunneling, and in particular relates to a method and system for identifying abnormal TBM tunneling parameter data. Background Technology
[0002] TBMs (hard rock tunnel boring machines) are crucial pieces of equipment used in hard rock tunnel construction. They offer advantages such as high tunneling efficiency, good safety, and environmental friendliness, and are widely used in tunnel excavation. During tunnel construction, the TBM collects relevant data such as thrust, torque, pressure, temperature, and current through sensors. This data is then converted from analog to digital and aggregated to a programmable logic controller (PLC). The PLC then uploads the data to the control host or via network to the cloud, forming the final TBM tunneling parameter data.
[0003] In recent years, research on TBM tunneling parameter data modeling, analysis and mining has received widespread attention. The proposed theoretical methods have been successfully applied in the fields of TBM tunneling performance prediction, tunneling parameter decision-making and surrounding rock condition identification, effectively improving the tunneling efficiency of TBM and providing theoretical and methodological support for intelligent TBM tunneling construction.
[0004] However, TBM operations are entirely open, with the equipment directly exposed to the construction site. The intense vibrations and impacts generated during the tunneling process, along with the high geothermal, high humidity, and high dust levels, cause interference in sensor data acquisition and signal transmission. Therefore, abnormal data within the tunneling parameters are unavoidable. The presence of abnormal TBM tunneling parameters can lead to deviations in subsequent data modeling, analysis, and mining results, and may even result in completely opposite predictions.
[0005] TBM tunneling involves numerous parameters, and traditional anomaly identification methods based on metrics such as data distance and spatial density suffer from the "curse of dimensionality," making them ineffective in identifying anomalies in TBM tunneling parameter data. Furthermore, TBM tunneling parameter data anomalies are more often manifested as abnormalities in the correlations between tunneling parameters. Summary of the Invention
[0006] To address the aforementioned technical problems, this invention provides a method and system for identifying abnormal TBM tunneling parameter data. It characterizes the nonlinear correlation between tunneling parameters using polynomial chaotic expansion (PCE), assigns membership functions to the tunneling parameter data based on data clustering, and determines whether the tunneling parameter data is abnormal based on the membership degree, thereby achieving effective identification of abnormal TBM tunneling parameter data.
[0007] This invention adopts the following technical solution: a method for identifying abnormal TBM tunneling parameter data, comprising the following steps:
[0008] Obtain the raw data of TBM tunneling parameters, and convert the raw data into tunneling parameter data in tabular form based on message writing rules;
[0009] By specifying the input and output tunneling parameters in the tunneling parameter data, a polynomial chaotic expansion regression model is constructed.
[0010] Given the clustering target number, construct the data clustering objective function by combining the polynomial chaotic expansion regression model;
[0011] The Lagrange multiplier method is used to optimize and solve the data clustering objective function to obtain the membership information of the tunneling parameter data.
[0012] Based on the membership information, and combined with the membership information judgment criteria, abnormal data in the tunneling parameter data are identified.
[0013] In a further embodiment, the tabular form of the tunneling parameter data is specifically represented as follows: each row represents a data point, and each column represents a parameter.
[0014] The conversion process of the tunneling parameter data in tabular form is as follows:
[0015] The original data is saved in the following format through text parsing: ;in, For root label, Indicates parameter name marker, For parameter names, Indicates parameter data markers, The parameter value;
[0016] All the separators in the data format are converted to spaces, and the tunneling parameter data is obtained by reading and organizing the data according to the table format.
[0017] In a further embodiment, the construction process of the polynomial chaotic expansion regression model is as follows:
[0018] Step 3.1: Based on the specified input and output tunneling parameters, establish a polynomial chaotic expansion regression model using polynomial chaotic expansion. ;
[0019] Step 3.2: Calculate the polynomial chaotic expansion regression model by minimizing the error between the model response and the polynomial chaotic expansion estimate. The polynomial chaotic expansion coefficients.
[0020] In a further embodiment, the process of constructing the data clustering objective function includes:
[0021] Let c be the given clustering target number. The data clustering objective function is expressed by the following formula:
[0022] ;
[0023] In the formula, This represents the objective function for data clustering. Represents the membership matrix. For sample size, Indicates the first Data points, Representing data points The output is the true value of the tunneling parameters. Indicates the first Multinomial chaotic expansion regression model for each subclass Indicates the first Data points of each subclass The predicted value, Representing data points For the Membership degree of each subclass, For membership degree of Exponentiation.
[0024] In a further embodiment, the optimization process for the data clustering objective function is as follows:
[0025] Step 5.1: Given training data Membership threshold and cluster target number c ;
[0026] Step 5.2: Randomly generate the initial membership matrix. In the initial membership matrix Select members with a membership degree greater than the membership degree threshold. The corresponding data is used as the actual training data for the multinomial chaotic expansion regression model. The actual training data is divided into training data subsets corresponding to the subclasses according to the clustering target number c.
[0027] Step 5.3: Check each training data subset to determine if it is an empty set. If it is an empty set, repeat step 5.2 to randomly generate a new membership matrix until each training data subset is not an empty set, and then proceed to step 5.4.
[0028] Step 5.4: Construct the first set of training data subsets based on non-empty sets. Polynomial chaotic expansion regression model for each subclass ;
[0029] Step 5.5: Calculate and update the membership degree using the Lagrange multiplier method. The specific formula is as follows:
[0030] ;
[0031] In the formula, Representing data points Credibility, , Data points representing the t-th cluster target number The predicted value;
[0032] Step 5.6: Based on the membership degree updated in Step 5.5, further update the current membership degree matrix until the iteration termination condition is met, resulting in the final membership degree matrix. This refers to membership information.
[0033] In a further embodiment, the method for identifying the abnormal data is as follows:
[0034] Based on the membership information, calculate the data points. The sum of the membership degrees and expressed as ,like This indicates that the corresponding data point is abnormal data. To determine the threshold.
[0035] In a further embodiment, the polynomial chaotic expansion regression model in step 3.1 The establishment process is as follows:
[0036] Define the input tunneling parameters as follows The output tunneling parameters are Based on input tunneling parameters and output tunneling parameters The correlation between them can be expressed as a multinomial chaotic expansion regression formula, as follows:
[0037] ;
[0038] In the formula, This represents a polynomial chaotic expansion regression model. This represents the order of the polynomial chaotic expansion. It is a polynomial and belongs to ,in, Represents the total order Not exceeding the given p polynomials of order 1 The dimension of the input tunneling parameters, yes The symbol for the set of dimensional nonnegative integers. Let be the coefficients of the polynomial chaotic expansion to be determined. Input tunneling parameters The tensor product of the univariate orthogonal polynomials corresponding to the polynomial chaotic expansion.
[0039] In a further embodiment, the calculation process of the polynomial chaotic expansion coefficients in step 3.2 includes:
[0040] Constructing matrix-form relationships: ;
[0041] In the formula, , Let be the coefficient vector of the polynomial chaotic expansion to be determined. Let P represent a real vector space of dimension P. P Indicates input tunneling parameters The maximum order; , Indicates output tunneling parameters The training sample vectors, Represents a real vector space of dimension N; , Let be a matrix, where each column contains the polynomial chaotic expansion estimate of N samples;
[0042] when At that time, the coefficient vector of the polynomial chaotic expansion is obtained by calculating the least squares regression. :
[0043] In the formula, For matrix permutations;
[0044] when Then, the problem of calculating the coefficients of the polynomial chaotic expansion is transformed into a problem of minimizing ℓ1:
[0045] In the formula, Denotes the ℓ1 norm, Denotes the ℓ2 norm, This represents the truncation error of the polynomial chaotic expansion.
[0046] In a further embodiment, the input tunneling parameters Maximum order P The calculation formula is as follows:
[0047] In the formula, The dimension of the input tunneling parameter vector, This represents the order of the polynomial chaotic expansion.
[0048] A TBM tunneling parameter anomaly data identification system, used to implement the TBM tunneling parameter anomaly data identification method as described above, includes:
[0049] The TBM tunneling parameter processing module is configured to acquire raw data of TBM tunneling parameters and convert the raw data into tunneling parameter data in tabular form based on message writing rules.
[0050] The polynomial chaotic expansion regression model construction module is configured to specify input tunneling parameters and output tunneling parameters in the tunneling parameter data to construct a polynomial chaotic expansion regression model.
[0051] The data clustering objective function construction module is set to construct the data clustering objective function in combination with the given number of clustering objectives and the polynomial chaotic expansion regression model.
[0052] The data clustering objective function optimization and solution module is configured to optimize and solve the data clustering objective function based on the data clustering objective function using the Lagrange multiplier method to obtain the membership information of the tunneling parameter data;
[0053] The abnormal data identification module is configured to identify abnormal data in the tunneling parameter data based on the membership information and the membership information judgment criteria.
[0054] This invention employs polynomial chaotic expansion to characterize the correlation between TBM tunneling parameters. It utilizes polynomial chaotic expansion and data clustering to identify anomalous TBM tunneling parameter data, solving the problem of difficulty in identifying anomalous TBM tunneling parameter data. This provides data support for TBM tunneling parameter data modeling, analysis, and mining. Compared to traditional linear models, it has better nonlinear characterization capabilities; compared to black-box models in machine learning, it can provide more reasonable and interpretable results; and compared to traditional metrics such as spatial distance and density, it can effectively avoid the "curse of dimensionality."
[0055] This invention constructs a data clustering objective function, establishes a strategy for solving the clustering objective function, obtains the membership degrees of TBM tunneling parameter data, and determines whether the tunneling parameter data is abnormal. Compared with traditional spectrum-based anomaly data identification methods, this invention can adapt to the low sampling rate of TBM tunneling parameter data and can provide more reasonable data partitioning results.
[0056] This invention achieves effective identification of abnormal TBM tunneling parameter data through a TBM tunneling parameter processing module, a multinomial chaotic expansion regression model construction module, a data clustering objective function construction module, a data clustering objective function optimization solution module, and an abnormal data identification module. Attached Figure Description
[0057] Figure 1This is a flowchart illustrating the TBM tunneling parameter anomaly data identification method of Example 1.
[0058] Figure 2 (a) is a diagram of the original data of the TBM tunneling parameters in Example 1.
[0059] Figure 2 (b) is a data format diagram of the text parsing in Example 1.
[0060] Figure 3 This is a diagram of the polynomial chaotic expansion regression model of Example 1.
[0061] Figure 4 This is a result diagram of the membership information of the final tunneling parameter data in Example 1.
[0062] Figure 5 This is a diagram showing the results of identifying abnormal TBM tunneling parameters in Example 1.
[0063] Figure 6 This is an architecture diagram of a TBM tunneling parameter anomaly data identification system in Example 2. Detailed Implementation
[0064] Embodiments of the present invention are described in detail below, examples of which are illustrated in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain the present invention, and should not be construed as limiting the present invention.
[0065] Example 1
[0066] like Figure 1 As shown in the figure, this embodiment discloses a method for identifying abnormal TBM tunneling parameter data, including the following steps:
[0067] The raw data of TBM tunneling parameters is obtained and converted into tabular tunneling parameter data based on message writing rules. In a further embodiment, the raw data of TBM tunneling parameters can be obtained through the data monitoring platform built into the TBM itself, or retrieved through the TBM cloud-edge collaborative monitoring platform. Correspondingly, the raw data in this embodiment contains at least 2000 sample data, such as data on 460 parameters including cutterhead speed, cutterhead thrust, cutterhead torque, propulsion system oil pressure, temperature, and motor current, including 200 abnormal data, such as... Figure 2 As shown in (a), the original data is in message format. Therefore, the message format data is converted into a table format, specifically: each row represents a data entry, and each column represents a parameter, as shown below. Figure 2 As shown in (b).
[0068] Input and output tunneling parameters are specified in the tunneling parameter data to construct a polynomial chaotic expansion regression model. In this embodiment, the propulsion speed is specified as the output tunneling parameter, and other tunneling parameters besides the propulsion speed are specified as input tunneling parameters. Based on this, a polynomial chaotic expansion regression model is constructed, referencing... Figure 3 .
[0069] Given a target number for clustering, a data clustering objective function is constructed using the polynomial chaotic expansion regression model. Specifically, the clustering objective function is built based on the error between the estimated and actual values of the polynomial chaotic expansion regression model of the advance speed, thus evaluating the current TBM tunneling parameter data clustering results.
[0070] Based on the aforementioned data clustering objective function, the Lagrange multiplier method is used to optimize and solve the data clustering objective function to obtain the membership information of the tunneling parameter data; in other words, the clustering objective function is updated iteratively by alternating between membership degree and data attribution information, and the iteration is terminated according to the iteration termination condition to obtain the final membership information of the tunneling parameter data, such as... Figure 4 As shown.
[0071] Based on the membership information and the membership information judgment criteria, abnormal data in the tunneling parameter data are identified, such as... Figure 5 As shown.
[0072] In a further embodiment, according to the above description, the message format is as follows: xml The process of converting the formatted data into tabular form of tunneling parameter data is as follows:
[0073] The original data is saved in the following format through text parsing: ;
[0074] in, For root label, Indicates parameter name marker, For parameter names, Indicates parameter data markers, The parameter value;
[0075] use txt The editing process involves converting all separators in the data format to spaces, reading and organizing the data according to the table format to obtain the tunneling parameter data. In other words, it involves sequentially reading each line of the message data, performing the corresponding conversions, and finally obtaining the data for the entire file.
[0076] In a further embodiment, the construction process of the polynomial chaotic expansion regression model is as follows:
[0077] Step 3.1: Based on the specified input and output tunneling parameters, establish a polynomial chaotic expansion regression model using polynomial chaotic expansion. ;
[0078] Step 3.2: Calculate the polynomial chaotic expansion regression model by minimizing the error between the model response and the polynomial chaotic expansion estimate. The polynomial chaotic expansion coefficients.
[0079] Furthermore, the polynomial chaotic expansion regression model in step 3.1 The establishment process is as follows:
[0080] Define the input tunneling parameters as follows The output tunneling parameters are Based on input tunneling parameters and output tunneling parameters The correlation between them can be expressed as a multinomial chaotic expansion regression formula, as follows:
[0081] ;
[0082] In the formula, This represents a polynomial chaotic expansion regression model. This represents the order of the polynomial chaotic expansion. It is a polynomial and belongs to ,in, Represents the total order Not exceeding the given p polynomials of order 1 The dimension of the input tunneling parameters, yes The symbol for the set of dimensional nonnegative integers. Let be the coefficients of the polynomial chaotic expansion to be determined. Input tunneling parameters The tensor product of the univariate orthogonal polynomials corresponding to the polynomial chaotic expansion.
[0083] Correspondingly, the calculation process of the polynomial chaotic expansion coefficients in step 3.2 includes:
[0084] Constructing matrix-form relationships: ;
[0085] In the formula, , Let be the coefficient vector of the polynomial chaotic expansion to be determined. Let P represent a real vector space of dimension P. P Indicates input tunneling parameters The maximum order; , Indicates output tunneling parameters The training sample vectors, Represents a real vector space of dimension N; , Let be a matrix, where each column contains the polynomial chaotic expansion estimate of N samples;
[0086] when At that time, the coefficient vector of the polynomial chaotic expansion is obtained by calculating the least squares regression. :
[0087] In the formula, For matrix permutations;
[0088] when Then, the problem of calculating the coefficients of the polynomial chaotic expansion is transformed into a problem of minimizing ℓ1:
[0089] In the formula, Denotes the ℓ1 norm, Denotes the ℓ2 norm, This represents the truncation error of the polynomial chaotic expansion.
[0090] It is worth mentioning that the input tunneling parameters Maximum order P The calculation formula is as follows:
[0091] In the formula, The dimension of the input tunneling parameter vector.
[0092] In another embodiment, the process of constructing the data clustering objective function includes:
[0093] Let c be the number of clustering targets. In this embodiment, c is 3. The data clustering objective function is expressed by the following formula:
[0094] ;
[0095] In the formula, This represents the objective function for data clustering. Represents the membership matrix. For sample size, Indicates the first Data points, Representing data points The output is the true value of the tunneling parameters. Indicates the first Multinomial chaotic expansion regression model for each subclass Indicates the first Data points of each subclass The predicted value, Representing data points For the Membership degree of each subclass, For membership degree of Exponentiation.
[0096] Among them, data points For the Membership degree of each subclass The following conditions must be met:
[0097] ;in, Indicates any, For data points The credibility of [the data] is calculated using the following formula:
[0098] ;
[0099] in, Represents the true value The minimum absolute error between the predicted value and the actual value. This represents the minimum absolute error between the actual value and the predicted value for any data point j.
[0100] It should be noted that, ;in, Representing data points The output is the true value of the tunneling parameters. Indicates the first Polynomial chaotic expansion regression model for each subclass For data points The predicted values of the output tunneling parameters, where the first... Polynomial chaotic expansion regression model for each subclass The method of obtaining it can be a polynomial chaotic expansion regression model Description of how to obtain it.
[0101] Based on the given data clustering objective function The Lagrange multiplier method is used to solve the clustering objective function and obtain the membership information of the tunneling parameter data. The steps include:
[0102] Step 5.1: Given training data Membership threshold and cluster target number c ;
[0103] Step 5.2: Randomly generate the initial membership matrix. In the initial membership matrix Select members with a membership degree greater than the membership degree threshold. The corresponding data is used as the actual training data for the multinomial chaotic expansion regression model. The actual training data is divided into training data subsets corresponding to the subclasses according to the clustering target number c.
[0104] Step 5.3: Check each training data subset to determine if it is an empty set. If it is an empty set, repeat step 5.2 to randomly generate a new membership matrix until each training data subset is not an empty set, and then proceed to step 5.4.
[0105] Step 5.4: Construct the first set of training data subsets based on non-empty sets. Polynomial chaotic expansion regression model for each subclass ;
[0106] Step 5.5: Calculate and update the membership degree using the Lagrange multiplier method. The specific formula is as follows:
[0107] ;
[0108] In the formula, Representing data points Credibility, , Data points representing the t-th cluster target number The predicted value.
[0109] Step 5.6: Based on the membership degree updated in Step 5.5, further update the current membership degree matrix until the iteration termination condition is met, resulting in the final membership degree matrix. This refers to membership information. The iteration termination condition can be a set maximum number of iterations, such as 100 iterations. In other embodiments, it can also be other iteration termination conditions.
[0110] Based on the above description, the method for identifying abnormal data in this embodiment is as follows:
[0111] Based on the membership information, calculate the data points. The sum of the membership degrees and expressed as ,like This indicates that the corresponding data point is abnormal data. To determine the threshold. This embodiment... The value is 0.8, combined with Figure 4 and 5 It can be seen that there are significant differences between abnormal data and normal data in terms of membership degree. Figure 5 The label 1 indicates an anomaly. It can be seen that the TBM tunneling parameter anomaly data identification method provided in this embodiment can accurately identify the anomaly data of the TBM tunneling parameters in this case.
[0112] In summary, based on both unprocessed and anomaly-removed TBM tunneling parameter data, a propulsion speed prediction model was established using polynomial chaotic expansion, resulting in a significant improvement in prediction accuracy. This embodiment addresses the difficulty in identifying anomaly data in TBM tunneling parameters, providing a reference for TBM condition identification, fault diagnosis, and subsequent data modeling, analysis, and mining.
[0113] Example 2
[0114] To implement the TBM tunneling parameter anomaly data identification method described in Example 1, this example discloses a TBM tunneling parameter anomaly data identification system, such as... Figure 6 The following are included:
[0115] The TBM tunneling parameter processing module is configured to acquire raw data of TBM tunneling parameters and convert the raw data into tunneling parameter data in tabular form based on message writing rules.
[0116] The polynomial chaotic expansion regression model construction module is configured to specify input tunneling parameters and output tunneling parameters in the tunneling parameter data to construct a polynomial chaotic expansion regression model.
[0117] The data clustering objective function construction module is set to construct the data clustering objective function in combination with the given number of clustering objectives and the polynomial chaotic expansion regression model.
[0118] The data clustering objective function optimization and solution module is configured to optimize and solve the data clustering objective function based on the data clustering objective function using the Lagrange multiplier method to obtain the membership information of the tunneling parameter data;
[0119] The abnormal data identification module is configured to identify abnormal data in the tunneling parameter data based on the membership information and the membership information judgment criteria.
[0120] It should be understood that although this specification is described according to various embodiments, not every embodiment contains only one independent technical solution. This way of describing the specification is only for clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment can also be appropriately combined to form other implementation methods that can be understood by those skilled in the art.
[0121] The detailed descriptions listed above are merely specific illustrations of feasible embodiments of the present invention and are not intended to limit the scope of protection of the present invention. All equivalent embodiments or modifications made without departing from the spirit of the present invention should be included within the scope of protection of the present invention.
Claims
1. A method for identifying abnormal TBM tunneling parameter data, characterized in that, Includes the following steps: The raw data of TBM tunneling parameters is obtained, and the raw data is converted into tabular tunneling parameter data based on message writing rules. The conversion process of the tabular tunneling parameter data is as follows: Through text parsing, the raw data is saved in the following data format: ;in, For root label, Indicates parameter name marker, For parameter names, Indicates parameter data markers, The parameter value; Convert all the separator symbols in the data format to spaces, and read and organize the tunneling parameter data according to the table format to obtain the tunneling parameter data; By specifying the input and output tunneling parameters in the tunneling parameter data, a polynomial chaotic expansion regression model is constructed. Given the target number of clustering, construct the data clustering objective function by combining the polynomial chaotic expansion regression model; The Lagrange multiplier method is used to optimize and solve the data clustering objective function to obtain the membership information of the tunneling parameter data. Based on the membership information and the membership information judgment criteria, abnormal data in the tunneling parameter data are identified. The construction process of the polynomial chaotic expansion regression model is as follows: Step 3.1: Based on the specified input and output tunneling parameters, establish a polynomial chaotic expansion regression model using polynomial chaotic expansion. ;Including: Defining the input tunneling parameters as The output tunneling parameters are Based on input tunneling parameters and output tunneling parameters The correlation between them can be expressed as a multinomial chaotic expansion regression formula, as follows: ； In the formula, This represents a polynomial chaotic expansion regression model. This represents the order of the polynomial chaotic expansion. It is a polynomial and belongs to ,in, Represents the total order Not exceeding the given p polynomials of order 1 The dimension of the input tunneling parameters, yes The symbol for the set of dimensional nonnegative integers. Let be the coefficients of the polynomial chaotic expansion to be determined. Input tunneling parameters The tensor product of the univariate orthogonal polynomials corresponding to the polynomial chaotic expansion; Step 3.2: Calculate the polynomial chaotic expansion regression model by minimizing the error between the model response and the polynomial chaotic expansion estimate. The polynomial chaotic expansion coefficients include: constructing matrix-form relations: ; In the formula, , Let be the coefficient vector of the polynomial chaotic expansion to be determined. Represents a real vector space of dimension P. P Indicates input tunneling parameters The maximum order; , Indicates output tunneling parameters The training sample vectors, Represents a real vector space of dimension N; , Let be a matrix, where each column contains the polynomial chaotic expansion estimate of N samples; when At that time, the coefficient vector of the polynomial chaotic expansion is obtained by calculating the least squares regression. : In the formula, For matrix permutations; when Then, the problem of calculating the coefficients of the polynomial chaotic expansion is transformed into a problem of minimizing ℓ1: In the formula, Denotes the ℓ1 norm, Denotes the ℓ2 norm, This represents the truncation error of the polynomial chaotic expansion.
2. The method for identifying abnormal TBM tunneling parameter data according to claim 1, characterized in that, The tabular form of the tunneling parameter data is as follows: each row represents one data point, and each column represents one parameter.
3. The method for identifying abnormal TBM tunneling parameter data according to claim 1, characterized in that, The process of constructing the data clustering objective function includes: Define the given clustering target number as c The data clustering objective function is expressed by the following formula: ； In the formula, This represents the objective function for data clustering. Represents the membership matrix. For sample size, Indicates the first Data points, Representing data points The output is the true value of the tunneling parameters. Indicates the first Multinomial chaotic expansion regression model for each subclass Indicates the first Data points of each subclass The predicted value, Representing data points For the Membership degree of each subclass, For membership degree of Exponentiation.
4. The method for identifying abnormal TBM tunneling parameter data according to claim 3, characterized in that, The optimization process for the data clustering objective function is as follows: Step 5.1: Given training data Membership threshold and cluster target number c ; Step 5.2: Randomly generate the initial membership matrix. In the initial membership matrix Select members with a membership degree greater than the membership degree threshold. The corresponding data is used as the actual training data for the multinomial chaotic expansion regression model. The actual training data is divided into training data subsets corresponding to the subclasses according to the clustering target number c. Step 5.3: Check each training data subset to determine if it is an empty set. If it is an empty set, repeat step 5.2 to randomly generate a new membership matrix until each training data subset is not an empty set, and then proceed to step 5.
4. Step 5.4: Construct the first set of training data subsets based on non-empty sets. Polynomial chaotic expansion regression model for each subclass ; Step 5.5: Calculate and update the membership degree using the Lagrange multiplier method. The specific formula is as follows: ； In the formula, Representing data points Credibility, , Data points representing the t-th cluster target number The predicted value; Step 5.6: Based on the membership degree updated in Step 5.5, further update the current membership degree matrix until the iteration termination condition is met, resulting in the final membership degree matrix. This refers to membership information.
5. The method for identifying abnormal TBM tunneling parameter data according to claim 1, characterized in that, The method for identifying the abnormal data is as follows: Based on the membership information, calculate the data points. The sum of the membership degrees and expressed as ,like This indicates that the corresponding data point is abnormal data. To determine the threshold.
6. The method for identifying abnormal TBM tunneling parameter data according to claim 1, characterized in that, The input tunneling parameters Maximum order P The calculation formula is as follows: ； In the formula, The dimension of the input tunneling parameter vector, This represents the order of the polynomial chaotic expansion.
7. A TBM tunneling parameter anomaly data identification system, used to implement the TBM tunneling parameter anomaly data identification method as described in any one of claims 1 to 6, characterized in that, include: The TBM tunneling parameter processing module is configured to acquire raw data of TBM tunneling parameters and convert the raw data into tunneling parameter data in tabular form based on message writing rules. The polynomial chaotic expansion regression model construction module is configured to specify input tunneling parameters and output tunneling parameters in the tunneling parameter data to construct a polynomial chaotic expansion regression model. The data clustering objective function construction module is set to construct the data clustering objective function in combination with the given number of clustering objectives and the polynomial chaotic expansion regression model. The data clustering objective function optimization and solution module is configured to optimize and solve the data clustering objective function based on the data clustering objective function using the Lagrange multiplier method to obtain the membership information of the tunneling parameter data; The abnormal data identification module is configured to identify abnormal data in the tunneling parameter data based on the membership information and the membership information judgment criteria.