Knowledge-data dual-driven tbm jamming risk intelligent early warning method and system

Abstract

The application provides a TBM jamming risk intelligent early warning method and system based on knowledge-data double driving, which comprises the following steps: acquiring multi-source parameter information collected in the TBM tunneling process of TBM tunneling cases under different stratum tunneling conditions, and constructing a multi-source parameter database; processing the data in the multi-source parameter database, and calculating data-driven jamming probability based on a robust time sequence anomaly detection model for abnormal changes; extracting expert experience knowledge based on TBM jamming literature research and experience knowledge in actual construction cases, and constructing a TBM tunneling jamming event knowledge base; determining the membership degree and the reliability of the fuzzy condition, constructing a TBM tunneling jamming event rule reasoning machine, and calculating knowledge-driven jamming probability based on a fuzzy reasoning machine model; multiplying the data-driven probability and the knowledge-driven probability by using a factor multiplication method, calculating the TBM tunneling jamming probability, and obtaining the TBM tunneling jamming risk prediction result based on the knowledge-data double driving.

Description

Technical Field

[0001] This invention belongs to the field of TBM tunnel excavation technology, and particularly relates to a TBM machine risk intelligent early warning method and system based on knowledge-data dual-drive. Background Technology

[0002] The statements in this section are merely background information related to the present invention and do not necessarily constitute prior art.

[0003] Tunnel construction methods include drill-and-blast, shield / TBM, cut-and-cover, and immersed tube methods, among which drill-and-blast and shield / TBM are the most widely used. For deep-buried, long, hard rock tunnels, the TBM method is the preferred construction method. With the continuous advancement of infrastructure construction in my country's northwest and southwest, a large number of deep-buried, long tunnels are under construction or planned, and TBM construction will be widely applied. In line with the national goal of intelligent tunnel construction, the next key development direction is to achieve seamless data flow, intelligent equipment interconnection, and multi-source information feedback throughout the tunnel construction process, maximizing the reduction of human labor in tunnel construction and realizing efficient, safe, and green tunnel construction.

[0004] For long, deeply buried tunnels, intelligent early warning of the risk of TBM jamming in the tunneling strata is crucial for the safety and efficiency of TBM construction, and is the primary task and a fundamental element for achieving intelligent TBM tunneling. However, due to the uncertainty of geological conditions and the difficulty in obtaining geological parameters in a timely manner during rapid TBM tunneling, the main operator often relies on subjective experience combined with changes in TBM tunneling parameters to drive the TBM, making it difficult to detect early warning signs of TBM jamming. Once a jamming accident occurs, it severely affects the TBM construction progress, greatly increases TBM construction costs, and may even cause casualties, seriously threatening tunnel construction safety.

[0005] In recent years, artificial intelligence technology has been developing rapidly, demonstrating good application results for complex nonlinear problems and spatiotemporal sequence problems. It can be used for intelligent prediction and early warning of TBM (Tunnel Boring Machine) jamming risks in strata. By mining big data and extracting expert experience and knowledge, a complete database and knowledge base can be built to continuously accumulate and develop TBM tunneling experience and technology, thereby reducing the occurrence of TBM jamming accidents in strata.

[0006] However, current artificial intelligence technology has several technical problems when used for SIM card risk warning: it relies too heavily on learning samples, and the quality of the samples directly affects the training and prediction results of the model; purely data-driven prediction models may overfit, and the model has poor generalization ability. Summary of the Invention

[0007] To overcome the shortcomings of the existing technologies, this invention provides a smart early warning method for TBM jamming risks based on a knowledge-data dual-drive approach. It constructs a complete database and knowledge base, continuously accumulates and develops TBM tunneling experience and technology, and realizes smart prediction and early warning of TBM jamming risks under complex geological conditions, thereby reducing the occurrence of TBM jamming accidents in tunneling formations.

[0008] To achieve the above objectives, one or more embodiments of the present invention provide the following technical solutions:

[0009] Firstly, a knowledge- and data-driven intelligent early warning method for TBM card machine risks is disclosed, including:

[0010] To acquire multi-source parameter information collected during the TBM tunneling process in TBM construction cases under different geological conditions, and to construct a multi-source parameter database;

[0011] The data in the multi-source parameter database is processed, including: obtaining a thresholdless recursive graph by expressing the features of the multi-source parameters based on the thresholdless recursive graph; performing feature encoding on the thresholdless recursive graph based on the extreme learning machine autoencoder; comparing the features of the normal tunneling feature graph and the encoded multi-dimensional time series thresholdless recursive graph of the current monitoring period to identify abnormal changes in the time series of multi-source parameters; and calculating the probability of machine jamming based on a robust time series anomaly detection model for abnormal changes.

[0012] Based on literature review of TBM tunneling machine jamming and experience from actual construction cases, we extract expert experience and construct a knowledge base for TBM tunneling machine jamming events.

[0013] Determine the membership degree and credibility of fuzzy conditions, construct a rule inference engine for TBM tunneling machine jam events, and calculate the knowledge-driven jam probability based on the fuzzy inference engine model;

[0014] The probability of TBM tunneling machine jamming is calculated by multiplying the data-driven probability and the knowledge-driven probability using the factor multiplication method, and the risk prediction results of TBM tunneling machine jamming based on knowledge-data dual-drive are obtained.

[0015] As a further technical solution, the multi-source parameter information includes TBM tunneling parameter information, geophysical drilling parameter information, surrounding rock parameter information, and TBM equipment parameter information;

[0016] The TBM tunneling parameters include cutterhead thrust, cutterhead torque, cutterhead rotation speed, penetration depth, net tunneling speed, and utilization rate.

[0017] The geophysical drilling parameters include drilling pressure, drilling torque, drilling speed, drilling rotation speed, seismic wave method parameters, and induced polarization method parameters;

[0018] The surrounding rock parameter information includes surrounding rock type, uniaxial compressive strength, rock mass integrity coefficient, rock quality index, Poisson's ratio, and elastic modulus;

[0019] The TBM equipment parameter information includes TBM type, excavation diameter, rated torque, rated thrust, support shoe pressure, and cutter configuration.

[0020] As a further technical solution, multi-source parameter feature representation is performed based on a thresholdless recursive graph, specifically as follows:

[0021] Second-level time series data of multi-source parameter information related to TBM machine accidents were extracted from a multi-source parameter database. The hyperparameters and thresholds of a robust time series anomaly detection model were adjusted and optimized. Then, the multi-dimensional time series data related to TBM machine accidents were converted into a two-dimensional threshold-free recursive graph through phase space transformation technology, thereby highlighting the dynamic characteristics of multi-source parameter data within the time series.

[0022] As a further technical solution, feature encoding is performed based on an extreme learning machine autoencoder, specifically including:

[0023] The thresholdless recursive graph, which represents the variation characteristics of multi-source parameters, is input into the autoencoder of the extreme learning machine. The autoencoder is used to quickly learn the image features of the thresholdless recursive graph, and the feature changes at different time periods are determined by image reconstruction.

[0024] As a further technical solution, identifying anomalous changes in multi-source parameter time series specifically includes:

[0025] The image features of the normal tunneling section of the TBM are stored in the self-pattern set to construct a TBM normal tunneling feature library. This library is used to compare with the multidimensional time series recursive graph features of the current monitoring period. If the internal characteristics of the multidimensional time series data of the current monitoring period change, the reconstructed recursive graph will show significant changes, which can be identified as abnormal changes.

[0026] As a further technical solution, a robust time-series anomaly detection model is used to calculate the data-driven card-handling probability, specifically including:

[0027] The reconstruction error value between the reconstructed thresholdless recursive graph and the original thresholdless recursive graph in the TBM normal tunneling graph library is calculated. Based on the reconstruction error value, the jam probability of the current set time series interval is calculated. Then, the average value of the jam probability of all time series intervals within the TBM tunneling length is calculated as the jam probability of the current TBM tunneling length, i.e., the data-driven jam probability.

[0028] As a further technical solution, the membership degree and credibility of fuzzy conditions are determined, specifically including:

[0029] A trapezoid is chosen as the membership function;

[0030] The credibility of knowledge rules in actual TBM malfunction events is divided into three categories: high, medium, and low. The credibility of each knowledge rule is calculated based on the support of itemsets from literature and engineering cases. Specifically, the number of literatures / cases containing the nth malfunction event knowledge rule is divided by the total number of literatures / cases. If the result is in the range [0.9,1], the credibility is high; if the result is in the range [0.8,0.9), the credibility is medium; and if the result is in the range [0.7,0.8), the credibility is low.

[0031] Using fuzzy rules in the knowledge base, and based on multi-source parameter data of the current time series interval, the Mamdani fuzzy inference method is used for inference.

[0032] As a further technical solution, the probability of knowledge-driven card activation is calculated based on a fuzzy inference engine model, specifically including:

[0033] The weighted average calculation method is used to calculate the membership degree of each knowledge rule in the current time series interval. The credibility of each knowledge rule is used as a weight factor and multiplied by the membership degree. The sum is then averaged to obtain the probability of machine jamming in the current set time series interval. Then, the average value of the probability of machine jamming in all time series intervals within the TBM tunneling progress is calculated as the probability of machine jamming in the current TBM tunneling progress, i.e., the knowledge-driven probability of machine jamming.

[0034] Secondly, a knowledge- and data-driven intelligent early warning system for TBM card machine risks was disclosed, including:

[0035] The multi-source parameter database construction module is configured to: acquire multi-source parameter information collected during the TBM tunneling process in TBM construction tunnel cases under different geological conditions, and construct a multi-source parameter database.

[0036] The data-driven machine jamming probability calculation module is configured to process data in a multi-source parameter database, including: obtaining a thresholdless recursive graph by expressing the features of multi-source parameters based on a thresholdless recursive graph; encoding the features of the thresholdless recursive graph based on an extreme learning machine autoencoder; comparing the features of the normal tunneling feature graph and the encoded multi-dimensional time series thresholdless recursive graph of the current monitoring period to identify abnormal changes in the time series of multi-source parameters; and calculating the data-driven machine jamming probability based on a robust time series anomaly detection model for abnormal changes.

[0037] The TBM tunneling machine jamming event knowledge base construction module is configured to: extract expert experience knowledge based on TBM jamming machine literature research and experience knowledge from actual construction cases, and construct a TBM tunneling machine jamming event knowledge base;

[0038] The knowledge-driven machine card probability calculation module is configured to: determine the membership degree and credibility of fuzzy conditions, construct a TBM tunneling machine card event rule inference engine, and calculate the knowledge-driven machine card probability based on the fuzzy inference engine model;

[0039] The TBM tunneling machine risk prediction module is configured to: multiply the data-driven probability and the knowledge-driven probability using a factor multiplication method to calculate the TBM tunneling machine probability and obtain the TBM tunneling machine risk prediction result based on knowledge-data dual-drive.

[0040] The above one or more technical solutions have the following beneficial effects:

[0041] This invention enables intelligent prediction and early warning of TBM jamming tendencies in deep-buried long tunnels through a dual knowledge- and data-driven approach. It solves the problems of poor robustness and generalization of purely data-driven models. By constructing a multi-source parameter database for TBMs and a knowledge base for tunneling jamming events, the model's adaptability to complex geological conditions is improved, and it has good scalability.

[0042] This invention addresses the problem in actual TBM tunnel construction where the main operator relies on subjective experience combined with changes in TBM tunneling parameters to determine whether the current tunneling stratum is stuck. This presents difficulties in detecting early warning signs of TBM jamming and results in low accuracy in jamming event identification. The invention introduces a knowledge-data dual-drive method, which solves the problem of low accuracy of a simple knowledge model, improves the accuracy of jamming event identification, and increases the time margin for jamming event processing.

[0043] Advantages of additional aspects of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0044] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0045] Figure 1 This is a flowchart illustrating the method of an embodiment of the present invention. Detailed Implementation

[0046] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0047] It should be noted that the terminology used herein is for the purpose of describing particular implementations only and is not intended to limit the exemplary implementations of the present invention.

[0048] Where there is no conflict, the embodiments and features in the embodiments of the present invention can be combined with each other.

[0049] Example 1

[0050] See appendix Figure 1 As shown, this embodiment discloses a TBM card machine risk intelligent early warning method based on knowledge-data dual-drive, including:

[0051] A. Collect multi-source parameter information collected during the TBM tunneling process from a large number of TBM construction tunnel cases under different geological conditions, preprocess and reduce noise of the data, and construct a multi-source parameter database;

[0052] B. Data-driven processing: Multi-source parameter feature representation is performed based on a thresholdless recursive graph, and feature encoding is performed based on an extreme learning machine autoencoder. Then, the feature encoding is input into a robust time series anomaly detection model to identify abnormal changes in the time series of multi-source parameters. The probability of data-driven machine failure is calculated based on the robust time series anomaly detection model.

[0053] C. Based on the literature review of TBM tunneling machine jamming machines and the experience knowledge from actual construction cases, extract expert experience knowledge and construct a TBM tunneling machine jamming machine event knowledge base.

[0054] D. Determine the membership degree and credibility of fuzzy conditions, construct a TBM tunneling machine jam event rule inference engine, and calculate the knowledge-driven jam probability based on the fuzzy inference engine model;

[0055] E. The data-driven probability and the knowledge-driven probability are multiplied by the factor multiplication method to calculate the TBM tunneling machine malfunction probability and obtain the TBM tunneling machine malfunction risk prediction result based on knowledge-data dual-driven.

[0056] F. When tunneling a new TBM tunnel, simply repeat steps A to E to achieve intelligent prediction and early warning of TBM tunneling machine jamming risks under complex geological conditions.

[0057] In step A above, the multi-source parameter information refers to the multi-source parameter information collected during the TBM tunneling process in a large number of TBM construction tunnel cases under different geological conditions. This includes TBM tunneling parameter information, geophysical drilling parameter information, surrounding rock parameter information, and TBM equipment parameter information.

[0058] TBM tunneling parameter information includes parameters such as cutterhead thrust, cutterhead torque, cutterhead speed, penetration depth, net tunneling speed, and utilization rate;

[0059] The drilling geophysical parameters include drilling pressure, drilling torque, drilling speed, drilling rotation speed, and parameters such as seismic wave method and induced polarization method.

[0060] The surrounding rock parameter information includes parameters such as surrounding rock type, uniaxial compressive strength, rock mass integrity coefficient, rock quality index, Poisson's ratio, and elastic modulus;

[0061] TBM equipment parameter information includes parameters such as TBM type, excavation diameter, rated torque, rated thrust, support shoe pressure, and cutter configuration.

[0062] The advantage of using the above information is its high correlation with TBM cascading failures, allowing it to reflect the probability of TBM cascading failures from multiple dimensions. Subsequent preprocessing yields standardized data samples, which are then used to construct a multi-source parameter database to support data-driven processing.

[0063] In step B above, the data-driven approach uses a robust time-series anomaly detection algorithm for modeling, learns the time-series change characteristics of multi-dimensional time series data, and realizes data-driven TBM card machine probability prediction based on the multi-source parameter information related to TBM card machines collected by the multi-source data module.

[0064] The robust time-series anomaly detection algorithm works by inputting time-series samples from a multi-source parameter database into the algorithm model, using multi-source parameters such as thrust, torque, and uniaxial compressive strength as input variables, and the jamming probability as the output variable. The algorithm automatically identifies abnormal changes in the time-series samples and obtains the jamming probability.

[0065] Specifically, this includes data preprocessing and noise reduction of multi-source parameter information. Data preprocessing employs missing value imputation and outlier removal; data noise reduction utilizes methods such as wavelet analysis, moving average, isolated forest, and local outlier factor analysis.

[0066] Second-level time series data of multi-source parameter information related to TBM jacking accidents are extracted from the preprocessed database. The hyperparameters and thresholds of the robust time series anomaly detection model are adjusted and optimized. Then, the multi-dimensional time series data related to TBM jacking accidents is converted into a two-dimensional threshold-free recursive graph through phase space transformation technology, thereby highlighting the dynamic characteristics of multi-source parameter data within the time series.

[0067] During extraction, the TBM equipment is equipped with a main control room, where the computer automatically records this data; it only needs to be exported. The second-level time series data specifically includes parameters such as cutterhead thrust, cutterhead torque, cutterhead rotation speed, penetration depth, net tunneling speed, and utilization rate.

[0068] When adjusting and optimizing the hyperparameters and thresholds of a robust time-series anomaly detection model, a pilot section pre-excavation study is typically conducted first. Based on the actual geological conditions revealed during the TBM excavation in the pilot section, the occurrence of machine jamming accidents is recorded. The hyperparameters and thresholds of the anomaly detection model are then corrected to ensure that the model's prediction results are consistent with the results revealed in the pilot section. The model can then be applied in practical engineering projects.

[0069] The advantage is that it can highlight the dynamic characteristics of multi-source parameter data within the time series.

[0070] Regarding the conversion of multidimensional time series data related to TBM machine accidents into two-dimensional thresholdless recursive graphs using phase space transformation technology, phase space reconstruction is essentially a process of reconstructing dynamical systems. It extends one-dimensional time series to a higher-dimensional phase space, thereby more effectively extracting the information contained within the original time series. Currently, the delay vector method is generally used for phase space reconstruction. The basic principle is that certain components in the system act on the evolution of an arbitrary component, and the change of that arbitrary component over time also carries the evolutionary information of other components. In other words, the entire dynamic change law of the system has a certain relationship with a certain arbitrary variable. The mathematical theoretical basis of phase space reconstruction technology is Takens' theorem. According to Takens' theorem, to ensure that the reconstructed phase space contains the characteristics of the original system, the dimension m of the embedded phase space and the attractor dimension d of the original system must satisfy: m ≥ 2d.

[0071] The thresholdless recursive graph, which represents the variation characteristics of multi-source parameters, is input into the autoencoder of the extreme learning machine. The autoencoder quickly learns the image features of the thresholdless recursive graph and determines whether the features have changed at different time periods through image reconstruction. The thresholdless recursive graph is then reconstructed to obtain a reconstructed thresholdless recursive graph.

[0072] Determining whether features have changed across time periods: This is automatically detected by the model, which uses a black-box network learned from samples to make this judgment. If a change occurs, the reconstructed thresholdless recursive graph will show a significant change, which can be identified as an abnormal change, i.e., a system freeze.

[0073] The image features of the normal tunneling section of the TBM are stored in the self-pattern set to construct a TBM normal tunneling feature library. This library is used to compare with the multidimensional time series recursive graph features of the current monitoring period. If the internal characteristics of the multidimensional time series data of the current monitoring period change, the reconstructed recursive graph will show significant changes, which can be identified as abnormal changes.

[0074] The reconstruction error between the reconstructed thresholdless recursive graph and the original thresholdless recursive graph in the TBM normal tunneling graph library is calculated. Based on the reconstruction error, the jam probability of the current set time series interval is calculated, typically using the mean squared error method. Then, the average jam probability of all time series intervals within the TBM tunneling length is calculated as the jam probability of the current TBM tunneling length, i.e., data-driven jam probability. For example, the average value can be used; if there are three intervals within the step length with probabilities of 0.8, 0.7, and 0.9, the average value is 0.8.

[0075] In step C above, a fuzzy inference engine method is used for modeling. Based on the literature review of TBM card machines and the experience knowledge from actual construction cases, expert experience knowledge is extracted to build a knowledge base and realize the diagnosis of TBM card machine events.

[0076] The use of fuzzy If-Then production rules ensures consistency with actual experience while also reflecting the fuzziness, uncertainty, and strength of knowledge rules.

[0077] In step D above, the membership degree and confidence level of the fuzzy conditions are determined. Actual TBM construction multi-source parameter data mostly follow a normal distribution; therefore, a trapezoidal function is chosen as the membership degree function. The confidence level of knowledge rules for actual TBM jam events can be divided into three categories: high, medium, and low. The confidence level of each knowledge rule is calculated based on the support of itemsets from literature and engineering cases. Specifically, the number of literature / cases containing the nth jam event knowledge rule is divided by the total number of literature / cases. Results in the interval [0.9, 1] indicate high confidence; results in the interval [0.8, 0.9) indicate medium confidence; and results in the interval [0.7, 0.8) indicate low confidence. Knowledge rules outside this interval are not included in the statistics to avoid rule redundancy and errors affecting prediction accuracy.

[0078] Using fuzzy rules in the knowledge base, and based on multi-source parameter data of the current time series interval, the Mamdani fuzzy inference method is used for inference.

[0079] The weighted average calculation method is used to calculate the membership degree of each knowledge rule in the current time series interval. The credibility of each knowledge rule is used as a weight factor and multiplied by the membership degree. The sum is then averaged to obtain the probability of machine jamming in the current set time series interval. Then, the average value of the probability of machine jamming in all time series intervals within the TBM tunneling progress is calculated as the probability of machine jamming in the current TBM tunneling progress, i.e., the knowledge-driven probability of machine jamming.

[0080] The above calculations can be performed using the binary comparison ranking method to determine the membership function and calculate the membership degree. The advantage is that it allows for the quantitative expression of qualitative knowledge.

[0081] In step E above, the TBM tunneling jam risk prediction result is calculated by factor multiplication, that is, multiplying the data-driven jam probability with the knowledge-driven jam probability. When the result is greater than 0.6, it can be determined that a jam accident has occurred in the current TBM tunneling progress.

[0082] Example 2

[0083] The purpose of this embodiment is to provide a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the above-described method.

[0084] Example 3

[0085] The purpose of this embodiment is to provide a computer-readable storage medium.

[0086] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs the steps of the above-described method.

[0087] Example 4

[0088] The purpose of this embodiment is to provide a knowledge- and data-driven intelligent early warning system for TBM card reader risks, including:

[0089] The multi-source parameter database construction module is configured to: acquire multi-source parameter information collected during the TBM tunneling process in TBM construction tunnel cases under different geological conditions, and construct a multi-source parameter database.

[0090] The data-driven machine jamming probability calculation module is configured to process data in a multi-source parameter database, including: obtaining a thresholdless recursive graph by expressing the features of multi-source parameters based on a thresholdless recursive graph; encoding the features of the thresholdless recursive graph based on an extreme learning machine autoencoder; comparing the features of the normal tunneling feature graph and the encoded multi-dimensional time series thresholdless recursive graph of the current monitoring period to identify abnormal changes in the time series of multi-source parameters; and calculating the data-driven machine jamming probability based on a robust time series anomaly detection model for abnormal changes.

[0091] The TBM tunneling machine jamming event knowledge base construction module is configured to: extract expert experience knowledge based on TBM jamming machine literature research and experience knowledge from actual construction cases, and construct a TBM tunneling machine jamming event knowledge base;

[0092] The knowledge-driven machine card probability calculation module is configured to: determine the membership degree and credibility of fuzzy conditions, construct a TBM tunneling machine card event rule inference engine, and calculate the knowledge-driven machine card probability based on the fuzzy inference engine model;

[0093] The TBM tunneling machine risk prediction module is configured to: multiply the data-driven probability and the knowledge-driven probability using the factor multiplication method to calculate the TBM tunneling machine probability, obtain the TBM tunneling machine risk prediction result based on knowledge-data dual-drive, and realize intelligent prediction and early warning of TBM tunneling machine risk under complex geological conditions.

[0094] The steps and methods involved in the apparatuses of Embodiments 2, 3, and 4 above correspond to those in Embodiment 1. For specific implementation details, please refer to the relevant description section of Embodiment 1. The term "computer-readable storage medium" should be understood as a single medium or multiple media including one or more instruction sets; it should also be understood as including any medium capable of storing, encoding, or carrying an instruction set for execution by a processor and enabling the processor to perform any of the methods in this invention.

[0095] Those skilled in the art will understand that the modules or steps of the present invention described above can be implemented using general-purpose computer devices. Optionally, they can be implemented using computer-executable program code, thereby allowing them to be stored in a storage device for execution by a computer device, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. The present invention is not limited to any particular combination of hardware and software.

[0096] While the specific embodiments of the present invention have been described above in conjunction with the accompanying drawings, this is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art without creative effort based on the technical solutions of the present invention are still within the scope of protection of the present invention.

Claims

1. A knowledge- and data-driven intelligent early warning method for TBM card machine risks, characterized by: include: To acquire multi-source parameter information collected during the TBM tunneling process in TBM construction cases under different geological conditions, and to construct a multi-source parameter database; The data in the multi-source parameter database is processed, including: obtaining a thresholdless recursive graph by expressing the features of the multi-source parameters based on the thresholdless recursive graph; performing feature encoding on the thresholdless recursive graph based on the extreme learning machine autoencoder; comparing the features of the normal tunneling feature graph and the encoded multi-dimensional time series thresholdless recursive graph of the current monitoring period to identify abnormal changes in the time series of multi-source parameters; and calculating the probability of machine jamming based on a robust time series anomaly detection model for abnormal changes. Multi-source parameter feature representation based on thresholdless recursive graphs is as follows: Second-level time series data of multi-source parameter information related to TBM machine accidents were extracted from a multi-source parameter database. The hyperparameters and thresholds of the robust time series anomaly detection model were adjusted and optimized. Then, the multi-dimensional time series data related to TBM machine accidents were converted into a two-dimensional threshold-free recursive graph through phase space transformation technology, thereby highlighting the dynamic characteristics of multi-source parameter data within the time series. Based on literature review of TBM tunneling machine jamming and experience from actual construction cases, we extract expert experience and construct a knowledge base for TBM tunneling machine jamming events. Determine the membership degree and credibility of fuzzy conditions, construct a rule inference engine for TBM tunneling machine jam events, and calculate the knowledge-driven jam probability based on the fuzzy inference engine model; Determining the membership degree and confidence level of fuzzy conditions specifically includes: A trapezoid is chosen as the membership function; The credibility of knowledge rules in actual TBM malfunction events is divided into three categories: high, medium, and low. The credibility of each knowledge rule is calculated based on the support of itemsets from literature and engineering cases. Specifically, the number of literatures / cases containing the nth malfunction event knowledge rule is divided by the total number of literatures / cases. If the result is in the range [0.9, 1], the credibility is high; if the result is in the range [0.8, 0.9), the credibility is medium; and if the result is in the range [0.7, 0.8), the credibility is low. Using fuzzy rules in the knowledge base, and based on multi-source parameter data of the current time series interval, the Mamdani fuzzy inference method is used for inference. The probability of TBM tunneling machine jamming is calculated by multiplying the data-driven probability and the knowledge-driven probability using the factor multiplication method, and the risk prediction results of TBM tunneling machine jamming based on knowledge-data dual-drive are obtained.

2. The TBM card machine risk intelligent early warning method based on knowledge-data dual-drive as described in claim 1, characterized in that, Feature encoding based on Extreme Learning Machine autoencoders specifically includes: The thresholdless recursive graph, which represents the variation characteristics of multi-source parameters, is input into the autoencoder of the extreme learning machine. The autoencoder is used to quickly learn the image features of the thresholdless recursive graph, and the feature changes at different time periods are determined by image reconstruction.

3. The TBM card machine risk intelligent early warning method based on knowledge-data dual-drive as described in claim 1, characterized in that, Identifying anomalous changes in multi-source parameter time series data specifically includes: The image features of the normal tunneling section of the TBM are stored in the self-pattern set to construct a TBM normal tunneling feature library. This library is used to compare with the multidimensional time series recursive graph features of the current monitoring period. If the internal characteristics of the multidimensional time series data of the current monitoring period change, the reconstructed recursive graph will show significant changes, which can be identified as abnormal changes.

4. The TBM card machine risk intelligent early warning method based on knowledge-data dual-drive as described in claim 1, characterized in that, The probability of a data-driven card being detected is calculated based on a robust time-series anomaly detection model, specifically including: The reconstruction error value between the reconstructed thresholdless recursive graph and the original thresholdless recursive graph in the TBM normal tunneling graph library is calculated. Based on the reconstruction error value, the jam probability of the current set time series interval is calculated. Then, the average value of the jam probability of all time series intervals within the TBM tunneling length is calculated as the jam probability of the current TBM tunneling length, i.e., the data-driven jam probability.

5. The TBM card machine risk intelligent early warning method based on knowledge-data dual-drive as described in claim 1, characterized in that, The probability of knowledge-driven card activation is calculated based on a fuzzy inference engine model, specifically including: The weighted average calculation method is used to calculate the membership degree of each knowledge rule in the current time series interval. The credibility of each knowledge rule is used as a weight factor and multiplied by the membership degree. The sum is then averaged to obtain the probability of machine jamming in the current set time series interval. Then, the average value of the probability of machine jamming in all time series intervals within the TBM tunneling progress is calculated as the probability of machine jamming in the current TBM tunneling progress, i.e., the knowledge-driven probability of machine jamming.

6. A knowledge-data dual-driven TBM card reader risk intelligent early warning system, employing the knowledge-data dual-driven TBM card reader risk intelligent early warning method as described in any one of claims 1-5, characterized in that, include: The multi-source parameter database construction module is configured to: acquire multi-source parameter information collected during the TBM tunneling process in TBM construction tunnel cases under different geological conditions, and construct a multi-source parameter database. The data-driven machine jamming probability calculation module is configured to process data in a multi-source parameter database, including: obtaining a thresholdless recursive graph by expressing the features of multi-source parameters based on a thresholdless recursive graph; encoding the features of the thresholdless recursive graph based on an extreme learning machine autoencoder; comparing the features of the normal tunneling feature graph and the encoded multi-dimensional time series thresholdless recursive graph of the current monitoring period to identify abnormal changes in the time series of multi-source parameters; and calculating the data-driven machine jamming probability based on a robust time series anomaly detection model for abnormal changes. The TBM tunneling machine jamming event knowledge base construction module is configured to: extract expert experience knowledge based on TBM jamming machine literature research and experience knowledge from actual construction cases, and construct a TBM tunneling machine jamming event knowledge base; The knowledge-driven machine card probability calculation module is configured to: determine the membership degree and credibility of fuzzy conditions, construct a TBM tunneling machine card event rule inference engine, and calculate the knowledge-driven machine card probability based on the fuzzy inference engine model; The TBM tunneling machine risk prediction module is configured to: multiply the data-driven probability and the knowledge-driven probability using a factor multiplication method to calculate the TBM tunneling machine probability and obtain the TBM tunneling machine risk prediction result based on knowledge-data dual-drive.

7. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the method described in any one of claims 1-5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it performs the steps of the method described in any one of claims 1-5.

Images