A risk assessment method and system for overseas high-speed railway urban tunnel construction

Abstract

The present invention relates to the field of risk assessment, and provides a method and system for risk assessment of urban tunnel construction in overseas high-speed railways. The risk assessment method includes the following steps: before tunnel construction, a risk plan is first formulated, risk factors are decomposed, a risk list is established, and an assessment mechanism is established for the dynamic changes of risk factors: the method for establishing the assessment mechanism includes: constructing a judgment matrix to obtain the weight of each factor, and at the same time performing dimensionality reduction and clustering on the risk factor matrix; obtaining a comprehensive weight based on the clustering and the constructed judgment matrix, and dynamically adjusting the comprehensive weight using a reinforcement learning method to calculate the risk assessment value. The risk assessment method for urban tunnel construction in overseas high-speed railways of the present invention proposes targeted risk control measures according to the risk level and risk type, which can effectively reduce the possibility and consequences of risk occurrence.

Description

Technical Field

[0001] The present invention relates to the field of risk assessment, and in particular to a method and system for risk assessment of overseas high-speed railway urban tunnel construction. Background Art

[0002] Traditional risk analysis methods mainly rely on expert experience and qualitative analysis, which makes it difficult to accurately quantify risks and to dynamically adjust according to changes in the construction process. In recent years, with the continuous development of risk management and tunnel engineering technology, some new risk analysis methods have emerged, such as: (1) Risk breakdown structure (RBS) method: decompose risk factors to form a hierarchical structure for easy identification and evaluation. (2) Analytic hierarchy process (AHP): through expert scoring, determine the weight of risk factors, and achieve quantitative assessment of risks. (3) Monte Carlo simulation method: by simulating the construction process, conduct probability analysis of risks and predict the possibility and consequences of risk occurrence.

[0003] However, the existing risk analysis methods still have the following shortcomings: (1) There is a lack of dedicated methods for overseas high-speed rail urban tunnel construction: the existing methods are difficult to adapt to the complex environment of overseas projects. (2) It is difficult to dynamically evaluate risks: the existing methods are difficult to dynamically adjust according to changes in the construction process. (3) It is difficult to formulate risk control measures: the existing methods are difficult to formulate effective control measures for different risk levels.

[0004] In view of this, the present invention is proposed. Summary of the invention

[0005] In view of this, the present invention proposes a risk assessment method for overseas high-speed rail urban tunnel construction. This method proposes a dynamic weight optimization system based on feature selection and clustering and reinforcement learning algorithm for the dynamic changes of risk factors during tunnel construction. The main risk features are extracted through PCA feature selection, K-Means clustering is used to simplify the grouping of risk factors, and reinforcement learning is used to dynamically adjust the risk weight according to the real-time feedback of the construction process, ultimately achieving efficient and accurate risk assessment, which is especially applicable to the situation where risk factors are constantly changing during tunnel construction.

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

[0007] The present invention provides a risk assessment method for overseas high-speed railway urban tunnel construction, comprising the following steps:

[0008] Before tunnel construction, a risk plan should be formulated, risk factors should be decomposed, a risk list should be established, and an evaluation mechanism should be established for the dynamic changes of risk factors:

[0009] The method for establishing the evaluation mechanism includes: constructing a judgment matrix to obtain the weight of each factor, and performing dimensionality reduction and clustering on the risk factor matrix;

[0010] A comprehensive weight is obtained based on the clustering and the constructed judgment matrix, and a risk assessment value is calculated after the comprehensive weight is dynamically adjusted using a reinforcement learning method.

[0011] Preferably, as a further feasible solution, a judgment matrix is ​​constructed to obtain the weight of each factor. After that, the consistency test is used to determine the CR value based on the CI and RI values:

[0012] ;

[0013] ;

[0014] in: - Maximum eigenvalue. If CR < 0.1, the matrix consistency is judged to be passed.

[0015] Preferably, as a further implementable solution, the method for performing dimensionality reduction and clustering on the risk factor matrix comprises the following steps: Perform eigendecomposition:

[0016] ;

[0017] Where: m-number of samples, which indicates the number of historical or real-time records collected during the risk assessment process; n-risk factor data, which indicates the number of dimensions of risk characteristics contained in each sample; is the eigencomponent in matrix decomposition;

[0018] Sort by singular value size, select the first d principal components whose cumulative variance contribution rate reaches 90-95%, and construct a dimensionality reduction matrix , and obtain the optimized risk factor matrix :

[0019] ;

[0020] After dimension reduction, the matrix Perform K-Means clustering to divide risk factors of the same type, level of impact and probability of occurrence into k Clusters.

[0021] Preferably, as a further implementable solution, the comprehensive weight is expressed by the following formula:

[0022] ;

[0023] in is the set of clustered risk factors of the 𝑘th class, For the k The number of risk factors in the class.

[0024] Preferably, as a further implementable solution, the method of dynamically adjusting the comprehensive weight using reinforcement learning includes: establishing a reinforcement learning framework, and then adjusting the strategy by introducing immediate rewards and Q values, and dynamically adjusting the comprehensive weight according to the adjusted strategy.

[0025] Preferably, as a further implementable solution, the method for establishing a reinforcement learning framework includes: ,action , and instant rewards The establishment of

[0026] The status It represents the current risk factor characteristics. The specific formula is:

[0027] ;

[0028] in: -The principal components extracted by PCA represent the simplified risk factor characteristics; -Comprehensive weight of K-Means clustering;

[0029] action Represents the current risk weight The specific formula for the adjustment operation is:

[0030] ;

[0031] in: -Fixed adjustment step size, indicating the magnitude of increasing or decreasing a certain weight;

[0032] Instant Rewards It is composed of the accuracy of risk assessment or the feedback value of tunnel construction safety.

[0033] Preferably, as a further feasible solution, the method of adjusting the strategy of the instant reward and the Q value comprises the following steps:

[0034] Update the Q value and compare the current Q value with the instant reward and the future maximum Q value Combined with adjustments, it gradually approaches the real long-term cumulative reward. The specific formula is as follows:

[0035] ;

[0036] in: - Current status and actions of Q Value, which indicates the cumulative reward that can be obtained by taking this action in this state, is initialized to zero; α -Learning rate, used to control Q The magnitude of the value update; γ - a discount factor, which indicates how much importance is placed on future rewards, ;

[0037] When the Q value gradually converges, stop training and output the optimized Q value and strategy ;

[0038] After each iteration, according to the final strategy Output Action , to dynamically adjust the comprehensive weight Get risk weight :

[0039] .

[0040] Preferably, as a further implementable solution, the method for calculating the risk assessment value is calculated based on the weight and contribution value of each risk factor.

[0041] In addition to providing a risk assessment method for overseas high-speed railway urban tunnel construction, the present invention also provides a risk assessment system for overseas high-speed railway urban tunnel construction, including:

[0042] Factor decomposition module: used to formulate a risk plan before tunnel construction, decompose risk factors, and establish a risk list:

[0043] Variable weight risk assessment module: used to establish an assessment mechanism for the dynamic changes of risk factors. The method for establishing the assessment mechanism includes: constructing a judgment matrix to obtain the weight of each factor, and at the same time performing dimensionality reduction and clustering on the risk factor matrix; obtaining a comprehensive weight based on the clustering and the constructed judgment matrix, and dynamically adjusting the comprehensive weight using a reinforcement learning method to calculate the risk assessment value.

[0044] In summary, the solution of the present invention has the following beneficial effects:

[0045] (1) Highly targeted: The present invention has established a comprehensive risk indicator system based on the risk characteristics of overseas high-speed railway urban tunnel construction, which can comprehensively identify and assess risks.

[0046] (2) Dynamic assessment: The present invention establishes a dynamic risk assessment model that can be dynamically adjusted according to changes in the construction process and identify new risks in a timely manner.

[0047] (3) Effective risk control measures: The present invention proposes targeted risk control measures based on risk levels and risk types, which can effectively reduce the possibility and consequences of risk occurrence.

[0048] (4) Strong operability: The present invention provides a detailed evaluation process and operation method, which is easy to promote and apply. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Various other advantages and benefits will become apparent to those skilled in the art by reading the following detailed description of the preferred embodiments. The accompanying drawings are only for the purpose of illustrating the preferred embodiments and are not to be considered as limiting the present invention. The same reference symbols are used throughout the accompanying drawings to represent the same components. In the accompanying drawings:

[0050] Figure 1 is a flow chart of the risk assessment method of the present invention;

[0051] Figure 2 A structural diagram of the risk list of the present invention;

[0052] Figure 3 is a block diagram of a risk assessment system of the present invention;

[0053] Figure 4 A flowchart of a computer device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0054] Here, exemplary embodiments will be described in detail, and examples thereof are shown in the accompanying drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present disclosure. On the contrary, they are only examples of devices and methods consistent with some aspects of the present disclosure.

[0055] The terms used in the present disclosure are only for the purpose of describing specific embodiments and are not intended to limit the present disclosure. The singular forms "a", "said" and "the" used in the present disclosure are also intended to include plural forms, unless the context clearly indicates otherwise. It should also be understood that the term "and / or" used herein refers to and includes any or all possible combinations of one or more associated listed items.

[0056] It should be understood that although the terms first, second, third, etc. may be used in the present disclosure to describe various information, such information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other. For example, without departing from the scope of the present disclosure, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information. Depending on the context, the word "if" used herein may be interpreted as "at the time" or "when" or "in response to determining",

[0057] Example

[0058] like Figure 1 As shown, the present invention provides a risk assessment method for overseas high-speed railway urban tunnel construction, comprising the following steps:

[0059] S1. Before tunnel construction, a risk plan should be developed, risk factors should be decomposed, and a risk list should be established;

[0060] S2. Establish an assessment mechanism for the dynamic changes of risk factors;

[0061] S21. The method for establishing the evaluation mechanism includes: constructing a judgment matrix to obtain the weight of each factor, and performing dimensionality reduction and clustering on the risk factor matrix, and obtaining a comprehensive weight according to the clustering and the constructed judgment matrix;

[0062] S22. After dynamically adjusting the comprehensive weight using the reinforcement learning method, the risk assessment value is calculated.

[0063] Specifically, in step S1, the steps to formulate a risk plan include:

[0064] (1) Determine the risk management objectives: ensure construction safety, protect the environment, control investment, guarantee construction period, and improve efficiency.

[0065] (2) Formulate risk management strategies: including risk avoidance, risk mitigation, risk retention and risk transfer strategies.

[0066] (3) Clarify organizational structure and responsibilities: Establish a risk management team and clarify the responsibilities and work content of each member.

[0067] (4) Establish a communication and coordination mechanism: establish an information sharing platform, hold risk management meetings regularly, and communicate risk information in a timely manner.

[0068] Specifically, after the risk plan is formulated, the risk factors are broken down using the risk breakdown structure (RBS) method to form a hierarchical structure, which makes it easier to identify and evaluate, and then a risk list is established.

[0069] Specifically, the methods for establishing a risk list include: natural risks, geological risks, technical risks and social risks, for example: Natural risks: rainstorms, floods, earthquakes, etc. Geological risks: groundwater, lithology, faults, weak strata, etc. Technical risks: construction methods, equipment selection, construction technology, etc. Social risks: land acquisition and demolition, social stability, environmental protection, etc. The risk factors identified above are classified and sorted to form a risk list, as follows: Figure 2 shown.

[0070] Specifically, step S2 is the main innovation of the solution of the present invention, which is to establish an evaluation mechanism for the dynamic changes of risk factors. It is mainly composed of steps S21 and S22. Step S21 includes step S211 to first construct a judgment matrix. The constructed judgment matrix is:

[0071] Comprehensive geological data, shield tunneling data and monitoring data of tunnel construction, use expert experience to make a preliminary weight allocation for risk factors, and n risk factors, construct a judgment matrix A =[ a ij ].

[0072] ;

[0073] in: a ij -factor i and j The relative importance of a ij =1 / a ji .

[0074] Next, in order to achieve reasonable judgment of the judgment matrix, after constructing the judgment matrix, the judgment matrix A =[ a ij ] is normalized to obtain the normalized judgment matrix .

[0075] ;

[0076] The normalized judgment matrix is ​​averaged row by row to obtain the weight of each factor. .

[0077] ;

[0078] Solve the CR value based on the CI and RI values ​​to determine whether its consistency is passed.

[0079] ;

[0080] ;

[0081] in: - Maximum eigenvalue.

[0082] If CR<0.1, the matrix consistency is judged to be passed, and finally, the weight vector matrix is ​​output. If the consistency judgment fails, the judgment matrix needs to be rebuilt.

[0083] Specifically, in step S21, including step S212, a series of processing is required on the risk factor matrix to better optimize and adjust the weights. The risk factor matrix here is different from the judgment matrix constructed previously. The series of processing here includes dimensionality reduction processing and clustering.

[0084] More preferably, the dimensionality reduction process is PCA dimensionality reduction:

[0085] Risk Factor Matrix Perform eigendecomposition:

[0086] ;

[0087] Where: m-number of samples, which indicates the number of historical or real-time records collected during the risk assessment process; n-risk factor data, which indicates the number of dimensions of risk characteristics contained in each sample; is the eigencomponent in matrix decomposition.

[0088] Sort by singular value (corresponding to the variance contribution of the principal component), select the first d principal components whose cumulative variance contribution rate reaches 90-95%, and construct a dimensionality reduction matrix , and obtain the optimized risk factor matrix .

[0089] ;

[0090] More preferably, the clustering is performed using K-Means clustering:

[0091] After dimension reduction, the matrix Perform K-Means clustering to divide risk factors of the same type, level of impact and probability of occurrence into k clusters, which means that the risk factors can be divided into k kind.

[0092] Specifically, step S21 also includes step S213, which is to obtain a comprehensive weight for each cluster according to the cluster and the constructed judgment matrix. , calculate the combined weight of all risk factors in this category .

[0093] ;

[0094] in: is the set of risk factors of the 𝑘th category, For the k The number of risk factors in the class.

[0095] Because the impact of many risk factors in tunnel engineering is interrelated, the clustering method can be used to group risk factors of the same type and the same level of impact and probability of occurrence, so as to evaluate the overall impact of each group of risks. After clustering, the weight of each risk factor is adjusted according to the impact of each group. In addition, feature selection and dimensionality reduction technology can help eliminate redundant or irrelevant risk factors, reduce unnecessary calculations, and improve the efficiency and prediction accuracy of the model. Therefore, in order to better adapt to the evaluation of variable risk factors in tunnel construction, the dimensionality reduction and clustering process is specially adopted.

[0096] Specifically, the method of dynamically adjusting the comprehensive weight by using reinforcement learning in step S22 is achieved by establishing a reinforcement learning framework, then adjusting the strategy by introducing immediate rewards and Q values, and dynamically adjusting the comprehensive weight according to the adjusted strategy.

[0097] Among them, step S22 includes step S221 to establish a reinforcement learning framework, and the specific method includes: state ,action , and instant rewards The establishment of

[0098] ① Status : Current risk factor characteristics (such as risk value, weight).

[0099] ;

[0100] in: -The principal components extracted by PCA represent the simplified risk factor characteristics; -The comprehensive weight of K-Means clustering, indicating the importance of risk factors after grouping.

[0101] ②Action : Current risk weight adjustment operation.

[0102] ;

[0103] in: -Fixed adjustment step, indicating the magnitude of increasing or decreasing a certain weight.

[0104] ③Rewards :It is composed of the accuracy of risk assessment and the feedback value of construction safety. After introducing the two factors of accurate or inaccurate risk assessment and safety during tunnel construction, it can provide more accurate and real-time feedback on tunnel risks.

[0105] The specific formula is as follows:

[0106] ;

[0107] ;

[0108] in: -The error part measures the difference between the predicted results of the risk assessment model and the actual risk results, introducing a negative sign to encourage action Reduce errors; -No. j The actual risk results of each sample (such as accident rate, loss amount, etc.); -Based on current risk weights The prediction results of the risk assessment model; m - Sample quantity, which indicates the number of monitoring data points during the construction process.

[0109] ;

[0110] - Safety reward part, designed based on real-time safety data during construction, measuring current actions Whether the safety of the construction environment has been improved; the reward for no accidents is positive R 1. When an accident occurs, a negative value will be penalized according to the severity and impact R 2.

[0111] The reason why reinforcement learning is used is that it is a learning method based on trial and error, which can gradually optimize decisions through interaction in a dynamic environment such as tunnel construction. In tunnel risk assessment, reinforcement learning can dynamically adjust weights based on real-time data feedback. For example, when the impact of a certain risk factor increases during tunnel construction, the reinforcement learning model can automatically increase the weight of the factor, thereby improving the accuracy of the overall risk assessment. Reinforcement learning optimizes the strategy through trial and error, indicating the probability of selecting action a in state s. Weight adjustment can be regarded as a continuous dynamic decision-making problem, so it has a good application in adjusting risk weights in the above risk assessment method.

[0112] Specifically, in step S22, including step S2222, the strategy is adjusted by introducing the instant reward and the Q value. The specific method includes the following steps:

[0113] Update the Q value and compare the current Q value with the instant reward and the future maximum Q value Combined with adjustments, it gradually approaches the real long-term cumulative reward. The specific formula is as follows:

[0114] ;

[0115] in: - Current status and actions of Q Value, which indicates the cumulative reward that can be obtained by taking this action in this state, is initialized to zero; α -Learning rate, used to control Q The magnitude of the value update; γ - a discount factor, which indicates how much importance is placed on future rewards, ;

[0116] When the Q value gradually converges, stop training and output the optimized Q value and strategy .

[0117] Strategy Indicates that in a given state Next select action The ultimate goal of reinforcement learning is to learn the optimal strategy, that is, The following choice can bring the greatest long-term cumulative reward Action The reinforcement learning algorithm selects the Q-Learning algorithm, and the model adjusts the strategy according to the Q value The Q value is updated by adding the current Q value to the immediate reward and the future maximum Q value Combined with adjustments, it gradually approaches the real long-term cumulative rewards.

[0118] Specifically, step S22 includes step S223 to dynamically adjust the comprehensive weight according to the adjusted strategy, which is actually after each round of iteration, according to the final strategy. Output Action , to dynamically adjust the comprehensive weight Get risk weight :

[0119] .

[0120] In short, through the training of the above Q-Learning reinforcement learning algorithm, the optimized dynamic risk weight is finally obtained. ,After iterations of reinforcement learning, the dynamic weights of risk assessment can timely reflect the risk changes during the construction process. The final risk weight matrix .

[0121] Specifically, the last step S22 also includes step S224 to calculate the risk assessment value, and the specific method includes:

[0122] In order to assess the risk level at different stages or monitoring points during tunnel construction, it is necessary to calculate a comprehensive risk assessment value based on the weight and contribution value of each risk factor. , and further divide them into different risk levels.

[0123] Contribution The contribution of each risk factor to the risk assessment at this stage is measured, reflecting the impact of the risk factor in the actual construction process. The contribution value is determined by the severity and impact of the risk (i.e., the weight). )Sure.

[0124] ;

[0125] in: - After standardization t The actual value of the risk factor, ; -No. t The weight of each risk factor.

[0126] After the subsequent risk assessment, there will be corresponding treatment mechanisms for different risk factors, including:

[0127] (1) Develop targeted risk control measures based on risk level and risk type, for example: Extremely high risk: Effective measures must be taken to avoid risk, otherwise the project cannot proceed. High risk: Effective measures should be taken to reduce risk and conduct risk monitoring. Medium risk: Effective measures should be taken to reduce risk and conduct risk monitoring. Low risk: No measures may be taken, but risk changes need to be monitored.

[0128] (2) Establish a risk warning mechanism: formulate risk warning indicators based on risk levels and risk change trends, and issue warning signals in a timely manner.

[0129] (3) Conduct dynamic monitoring: regularly collect construction process information, update risk assessment results, and adjust risk control measures in a timely manner.

[0130] (4) After the project is completed, the effectiveness of risk management work will be evaluated, lessons learned will be summarized, and closed-loop management will be formed. The evaluation content includes: the implementation of risk control measures, the occurrence of risks, the cost-effectiveness of risk management work, etc. Based on the evaluation results, the risk management process and methods will be improved to improve the risk management level.

[0131] In addition, the present invention provides a risk assessment method and a risk assessment system. Figure 3 As shown, including:

[0132] Factor decomposition module 101: It is used to formulate a risk plan before tunnel construction, decompose risk factors, and establish a risk list:

[0133] Variable-weighted risk assessment module 102: used to establish an assessment mechanism for the dynamic changes of risk factors. The method for establishing the assessment mechanism includes: constructing a judgment matrix to obtain the weight of each factor, and performing dimensionality reduction and clustering on the risk factor matrix; obtaining a comprehensive weight based on the clustering and the constructed judgment matrix, and dynamically adjusting the comprehensive weight using a reinforcement learning method to calculate the risk assessment value.

[0134] In summary, the risk assessment method and system of the present invention solve the following problems: (1) The overseas project environment is complex and the risk factors are diverse: Overseas projects usually face complex geological conditions, hydrological environment, social and cultural background, etc., which makes it difficult to fully identify and assess risk factors. (2) Risk control measures are difficult to formulate: Due to the diversity and interrelationship of risk factors, traditional risk analysis methods are difficult to effectively formulate targeted control measures. (3) Risk management work is difficult to dynamically adjust: There are many uncertainties in the tunnel construction process, and traditional risk management work is difficult to dynamically adjust according to actual conditions.

[0135] In specific implementation, the above modules can be implemented as independent entities, or can be arbitrarily combined and implemented as the same or several entities. The specific implementation of the above units can refer to the previous method embodiments, which will not be repeated here.

[0136] Figure 4 This is a schematic diagram of the structure of a computer device disclosed in the present invention. Figure 4 As shown, the computer device 400 includes at least a memory 402 and a processor 401; the memory 402 is connected to the processor via a communication bus 403, and is used to store computer instructions executable by the processor 401, and the processor 401 is used to read computer instructions from the memory 402 to implement the steps of the method described in any of the above embodiments.

[0137] For the above-mentioned device embodiments, since they basically correspond to the method embodiments, the relevant parts can refer to the partial description of the method embodiments. The device embodiments described above are only schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the disclosed solution. A person of ordinary skill in the art can understand and implement it without paying any creative work.

[0138] Computer readable media suitable for storing computer program instructions and data include all forms of non-volatile memory, media and memory devices, including, for example, semiconductor memory devices (e.g., EPROM, EEPROM and flash memory devices), magnetic disks (e.g., internal disks or removable disks), magneto-optical disks, and CD ROM and DVD-ROM disks. The processor and memory may be supplemented by, or incorporated in, special purpose logic circuits.

[0139] Finally, it should be noted that although this specification contains many specific implementation details, these should not be interpreted as limiting the scope of any invention or the scope of protection claimed, but are mainly used to describe the features of specific embodiments of specific inventions. Certain features described in multiple embodiments within this specification may also be implemented in combination in a single embodiment. On the other hand, the various features described in a single embodiment may also be implemented separately in multiple embodiments or in any suitable sub-combination. In addition, although features may work in certain combinations as described above and even initially claimed as such, one or more features from the claimed combination may be removed from the combination in some cases, and the claimed combination may point to a sub-combination or a variation of a sub-combination.

[0140] Similarly, although operations are depicted in a particular order in the accompanying drawings, this should not be understood as requiring that these operations be performed in the particular order shown or performed sequentially, or requiring that all illustrated operations be performed to achieve the desired results. In some cases, multitasking and parallel processing may be advantageous. In addition, the separation of various system modules and components in the above-described embodiments should not be understood as requiring such separation in all embodiments, and it should be understood that the described program components and systems can generally be integrated together in a single software product, or packaged into multiple software products.

[0141] Thus, specific embodiments of the subject matter have been described. In some cases, the actions described in this disclosure can be performed in a different order and still achieve the desired results. In addition, the processing depicted in the accompanying drawings does not necessarily require the specific order or sequential order shown to achieve the desired results. In some implementations, multitasking and parallel processing may be advantageous.

[0142] The above description is only a preferred embodiment of the present disclosure and is not intended to limit the present disclosure. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present disclosure shall be included in the scope of protection of the present disclosure.

Claims

1. A risk assessment method for overseas high-speed railway urban tunnel construction, characterized in that: The steps include: Before tunnel construction, a risk plan should be formulated, risk factors should be decomposed, a risk list should be established, and an evaluation mechanism should be established for the dynamic changes of risk factors: The method for establishing the evaluation mechanism includes: constructing a judgment matrix to obtain the weight of each factor, and performing dimensionality reduction and clustering on the risk factor matrix; Obtaining a comprehensive weight based on the clustering and the constructed judgment matrix, dynamically adjusting the comprehensive weight using a reinforcement learning method, and calculating a risk assessment value; Construct a judgment matrix to obtain the weight of each factor After that, the consistency test is used to determine the CR value based on the CI and RI values: in: - Maximum eigenvalue, if CR < 0.1, the matrix consistency is judged to be passed; The method for performing dimensionality reduction and clustering on the risk factor matrix comprises the following steps: Perform eigendecomposition: Where: m-number of samples, which indicates the number of historical or real-time records collected during the risk assessment process; n-risk factor data, which indicates the number of dimensions of risk characteristics contained in each sample; is the eigencomponent in matrix decomposition; Sort by singular value size, select the first d principal components whose cumulative variance contribution rate reaches 90-95%, and construct a dimensionality reduction matrix , and obtain the optimized risk factor matrix : After dimension reduction, the matrix Perform K-Means clustering to group similar risk factors into k clusters; The comprehensive weight is expressed by the following formula: in is the set of clustered risk factors of the 𝑘th class, For the k the number of risk factors in the class; The method of dynamically adjusting the comprehensive weight by using the reinforcement learning method includes: establishing a reinforcement learning framework, and then adjusting the strategy by introducing instant rewards and Q values, and dynamically adjusting the comprehensive weight according to the adjusted strategy; Methods for building reinforcement learning frameworks include: ,action , and instant rewards The establishment of The status It represents the current risk factor characteristics. The specific formula is: in: -The principal components extracted by PCA represent the simplified risk factor characteristics; -Comprehensive weight of K-Means clustering; action Represents the current risk weight The specific formula for the adjustment operation is: in: -Fixed adjustment step size, indicating the magnitude of increasing or decreasing a certain weight; Instant Rewards It is composed of the accuracy of risk assessment and the feedback value of tunnel construction safety.

2. The risk assessment method according to claim 1, characterized in that: The method of adjusting the strategy by using instant rewards and Q values ​​includes the following steps: Update the Q value and compare the current Q value with the instant reward and the future maximum Q value Combined with adjustments, it gradually approaches the real long-term cumulative reward. The specific formula is as follows: in: - Current status and actions of Q Value, which indicates the cumulative reward that can be obtained by taking this action in this state, is initialized to zero; α -Learning rate, used to control Q The magnitude of the value update; γ - a discount factor, which indicates how much importance is placed on future rewards, ; When the Q value gradually converges, stop training and output the optimized Q value and strategy ; After each iteration, according to the final strategy Output Action , to dynamically adjust the comprehensive weight Get risk weight : 。 3. The risk assessment method according to any one of claims 1-2, characterized in that: The method for calculating the risk assessment value is based on the weight and contribution value of each risk factor.

4. A risk assessment system using the risk assessment method according to any one of claims 1 to 3, characterized in that: include: Factor decomposition module: used to formulate a risk plan before tunnel construction, decompose risk factors, and establish a risk list: Variable weight risk assessment module: used to establish an assessment mechanism for the dynamic changes of risk factors. The establishment method of the assessment mechanism includes: constructing a judgment matrix to obtain the weight of each factor, and performing dimensionality reduction and clustering on the risk factor matrix; A comprehensive weight is obtained based on the clustering and the constructed judgment matrix, and a risk assessment value is calculated after the comprehensive weight is dynamically adjusted using a reinforcement learning method.

5. 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, the steps of the risk assessment method according to any one of claims 1 to 2 are implemented.

Images