Civil engineering structure analysis system and program

Abstract

To provide a system and program for analyzing civil engineering structures that shortens the time required to present analysis results to a user. [Solution] The civil engineering structure analysis system 10 comprises a web server 20 that provides a numerical analysis site for accepting data necessary for the numerical analysis of civil engineering structures, a cloud database 30 in which the data accepted at the numerical analysis site is stored, and an analysis device 50 that performs numerical analysis of civil engineering structures based on the data stored in the cloud database 30 and stores the results in the cloud database 30.

Description

[Technical Field]

[0001] The present invention relates to a system and program for analyzing civil engineering structures. [Background technology]

[0002] Patent Document 1 discloses a stress monitoring device. This stress monitoring device has an information acquisition unit that acquires measurement information of deformation or strain on structural members that make up a structure, and a storage unit that stores linked stress information that links the measurement information to design information of the structure and arrangement information of the structural members, including arrangement position information of the structural members that is linked to the design information, and is capable of outputting the linked stress information to a terminal. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] International Publication No. 2021 / 010263 Summary of the Invention [Problem to be solved by the invention]

[0004] Generally, in order to conduct numerical analysis of civil engineering structures, repeated prior meetings are required between the user and the analysis company that provides the numerical analysis service, and it takes time for the user to obtain the analysis results. Furthermore, during the analysis process, the analysis company's operator must repeatedly perform trial calculations, which not only takes time but also places a heavy workload on the analysis company. An object of the present invention is to provide a system and program for analyzing civil engineering structures that shortens the time required to present analysis results to the user. [Means for solving the problem]

[0005] The invention described in claim 1 is a civil engineering structure analysis system comprising: a web server that provides a numerical analysis site for receiving data required for the numerical analysis of civil engineering structures from users; a cloud database that stores the data received at the numerical analysis site; and an analysis device that performs numerical analysis of civil engineering structures based on the data stored in the cloud database and stores the results in the cloud database, wherein the analysis device comprises a database construction unit that constructs the results of the numerical analysis as an analysis result database; an evaluation unit that confirms that the accuracy of a machine learning model trained using data from the analysis result database meets a predetermined evaluation standard; and a reverse analysis is performed using the trained machine learning model that has been confirmed by the evaluation unit to meet the evaluation standard, to estimate analytical physical property values ​​of the civil engineering structures, and to use the analytical physical property values ​​to carry out an analysis. Then, a forward analysis was performed and the obtained This civil engineering structure analysis system includes an analytical physical property value estimation unit that determines the degree of agreement between a numerical analysis result and measurement data corresponding to the numerical analysis result, and an analysis unit that performs numerical analysis using the analytical physical property value estimated by the analytical physical property value estimation unit if the degree of agreement satisfies a predetermined judgment criterion, and if the degree of agreement does not satisfy the judgment criterion, the database construction unit adds the numerical analysis result to the analysis result database and retrains the machine learning model.

[0006] The invention described in claim 2 is the civil engineering structure analysis system described in claim 1, in which the analysis device confirms that the data stored in the cloud database has been updated at a predetermined interval.

[0007] The invention described in claim 3 is the civil engineering structure analysis system described in claim 2, wherein the learning algorithm of the machine learning model is LightGBM.

[0008] The invention described in claim 4 includes a web server that provides a numerical analysis site for receiving data required for the numerical analysis of civil engineering structures from users, and a data storage unit that stores the data received at the numerical analysis site. The aforementionedAn analysis system for civil engineering structures, comprising: an analysis device that performs a numerical analysis of a civil engineering structure based on data stored in a data storage unit and stores the results in the data storage unit, wherein the analysis device comprises a database construction unit that constructs the results of the numerical analysis as an analysis result database; an evaluation unit that confirms that the accuracy of a machine learning model trained using data from the analysis result database meets a predetermined evaluation standard; and a back analysis unit that performs a back analysis using the trained machine learning model that has been confirmed by the evaluation unit to meet the evaluation standard, estimates analytical physical property values ​​of the civil engineering structure, and uses the analytical physical property values ​​to perform an analysis. Then, a forward analysis was performed and the obtained This civil engineering structure analysis system includes an analytical physical property value estimation unit that determines the degree of agreement between a numerical analysis result and measurement data corresponding to the numerical analysis result, and an analysis unit that, if the degree of agreement satisfies a predetermined judgment criterion, performs numerical analysis at a predetermined position using the analytical physical property value estimated by the analytical physical property value estimation unit, and, if the degree of agreement does not satisfy the judgment criterion, the database construction unit adds the numerical analysis result to the analysis result database and retrains the machine learning model.

[0009] The invention described in claim 5 provides an analysis device connected to a web server that provides a numerical analysis site for receiving data required for the numerical analysis of civil engineering structures from users and a data storage unit that stores the data received at the numerical analysis site, and that performs numerical analysis of civil engineering structures based on the data stored in the data storage unit and stores the results in the data storage unit, the analysis device comprising: database construction means that constructs the results of the numerical analysis as an analysis result database; evaluation means that confirms that the accuracy of a machine learning model that has been trained using data from the analysis result database as training data satisfies a predetermined evaluation standard; and a back analysis that uses the trained machine learning model that has been confirmed by the evaluation means to satisfy the evaluation standard, to estimate analytical physical property values ​​of the civil engineering structure, and Then, a forward analysis was performed and the obtainedThe program includes an analytical property value estimation means that determines the degree of agreement between a numerical analysis result and measurement data corresponding to the numerical analysis result, and when the degree of agreement satisfies a predetermined criterion, functions as an analysis means that performs numerical analysis at a predetermined position using the analytical property value estimated by the analytical property value estimation means, and when the degree of agreement does not satisfy the criterion, the database construction means adds the numerical analysis result to the analysis result database and re-trains the machine learning model. [Effects of the Invention]

[0010] According to the present invention, it is possible to provide a system and program for analyzing civil engineering structures that shortens the time required to present analysis results to the user. [Brief explanation of the drawings]

[0011] [Figure 1] 1 is an explanatory diagram showing the configuration of a tunnel analysis system according to an embodiment of the present invention; [Figure 2] 10 is a flowchart showing the operation of the tunnel analysis system. [Figure 3] This is an explanatory diagram showing an example of numerical analysis using the tunnel analysis system. [Figure 4] FIG. 4 is a plan view showing the positions of the house in the numerical analysis example shown in FIG. 3. [Figure 5] 10 is a graph showing an example of an analysis result obtained by an analytical property value estimation unit of an analysis device provided in the tunnel analysis system. DETAILED DESCRIPTION OF THE INVENTION

[0012] Next, embodiments of the present invention will be described with reference to the accompanying drawings to facilitate understanding of the present invention. Note that in the drawings, parts that are not relevant to the description may be omitted.

[0013] A tunnel analysis system (an example of a civil engineering structure analysis system) 10 according to an embodiment of the present invention can provide the results of numerical analysis of a tunnel (an example of a civil engineering structure) by an analysis company based on user operations. The civil engineering structure is not limited to a tunnel, but may be any civil engineering structure whose behavior needs to be simulated during construction, such as a slope, an embankment, a cut, a dam, or a bridge. Hereinafter, "numerical analysis" may be simply referred to as "analysis."

[0014] As shown in FIG. 1, the tunnel analysis system 10 includes a service-providing web server 20, a cloud database 30, and an analysis device 50. The service-providing web server 20, the cloud database 30, and the analysis device 50 are connected to one another via the Internet.

[0015] The service-providing web server 20 can provide a numerical analysis site (web application) for accepting services for numerically analyzing tunnels. This numerical analysis site serves as an interface for analysis services provided by analysis businesses, and can accept tunnel analysis requests and data required for the analysis from user terminals 12 used by users. The data required for analysis includes measurement data at specified measurement points in the tunnel and types of analytical models. Note that multiple analytical models are prepared depending on, for example, the tunnel shape, the condition of the surrounding ground, the shape of the mountain, etc.

[0016] The cloud database (an example of a data storage unit) 30 is a database constructed on a cloud environment, and stores data received at the numerical analysis site. Additionally, the cloud database 30 stores the analysis results and other data sent from the analysis device 50, and users can check this data via the numerical analysis site. Instead of the cloud database 30, any data storage means capable of storing data may be used.

[0017] Analysis device 50 monitors cloud database 30 at a predetermined cycle, and when the data stored in cloud database 30 is updated, it performs numerical analysis of the tunnel based on the updated data and stores the analysis results in cloud database 30. This predetermined cycle is, for example, 60 to 300 seconds, and data updates also include the creation of new data. Analysis device 50 has database construction unit 502, evaluation unit 504, analytical physical property value estimation unit 506, and analysis unit 508. Database construction unit 502, evaluation unit 504, analytical physical property value estimation unit 506, and analysis unit 508 are each realized by a program executed by a CPU built into analysis device 50.

[0018] Database construction unit (an example of a database construction means) 502 can accumulate the results of analysis by analysis device 50 for each analysis model and construct an analysis result database. The constructed analysis result database is used as training data for the machine learning model. The data constituting the analysis result database includes, for example, coordinate data, displacement data, stress data, and strain data.

[0019] The evaluation unit 504 (an example of an evaluation means) can confirm that the accuracy of the machine learning model trained using the data in the analysis result database as training data satisfies a predetermined evaluation standard. Here, the machine learning model is a model for estimating analytical physical properties of the ground through which a tunnel is excavated, and will be described in detail later. The evaluation unit 504 compares the analytical values ​​based on the analytical physical property values ​​calculated using the trained machine learning model with the measurement values ​​corresponding to these analytical values, and confirms that the analytical values ​​satisfy a predetermined evaluation criterion, such as whether the root mean square error between these values ​​is equal to or less than a predetermined value.

[0020] The analytical physical property value estimation unit (an example of analytical physical property value estimation means) 506 can estimate analytical physical property values ​​(e.g., deformation coefficient and adhesion) of the tunnel by inverse analysis based on the measurement data stored in the cloud database 30 and the first analytical model selected by the user. The estimation of analytical physical property values ​​(inverse analysis) is performed by a trained machine learning model that has been confirmed by the evaluation unit 504 to satisfy the evaluation criteria. In addition, the analytical property value estimation unit 506 can determine the analytical value at the measurement point based on the estimated analytical property value, and compare the determined analytical value (analysis result) with the measurement data corresponding to this analytical value to determine the degree of agreement. If the analysis property value estimation unit 506 compares the analysis value with the measurement data and finds that the degree of match satisfies a predetermined criterion (if it is determined that there is a substantial match), the estimated analysis property value is stored in the cloud database 30, and if the degree of match does not satisfy the predetermined criterion (if it is determined that there is no substantial match), the database construction unit 502 adds the analysis result to the analysis result database.

[0021] The analysis unit (an example of an analysis means) 508 can perform analysis of the area ahead of the face by, for example, direct analysis using the finite difference method or the finite element method, using the analysis property values ​​estimated by the analysis property value estimation unit 506 and a second analysis model selected by the user. In addition, specific examples of data obtained as a result of this analysis include the physical properties of the ground ahead of the tunnel face, and the safety and construction management standard values ​​at the position to be evaluated ahead of the tunnel face. The second analysis model may be the same as the first analysis model. The analysis results are stored in the cloud database 30 and are shown to the user via the numerical analysis site.

[0022] Next, the operation of the tunnel analysis system 10 (a method for analyzing a civil engineering structure) will be described with reference to FIG. The tunnel analysis is carried out according to the following steps S1 to S12: executionHowever, if possible, the steps may be executed interchangeably or in parallel. In Fig. 2, steps S5 to S8 and S11 are processes that do not depend on an operator (person). To facilitate understanding, the following explanation will be based on a hypothetical example shown in Figure 3. In this hypothetical example, a house is located directly above a tunnel that is being excavated through a ground consisting of geology A and the underlying geology B, and there is concern that the house may subside due to ground deformation caused by the tunnel excavation. Because there are limits to what can be achieved by preliminary surveys before tunnel excavation, it is necessary to estimate the geological condition of the ground through back analysis once the tunnel has been excavated to a certain extent and to investigate the impact on the house. Point B to be evaluated is set, in detail, at positions P1 to P5 around the house as shown in Figure 4, for example.

[0023] (Step S1) After organizing the analysis conditions, the user selects the first analysis model corresponding to the tunnel being excavated on the numerical analysis site. The content selected in step S1 is stored in the cloud database 30.

[0024] (Step S2) The user inputs the data required for analysis. The data required for analysis includes measurement data at the target location (point A, which is the measurement point) shown in Figure 3. The target location is located a predetermined distance forward and upward from the vicinity of the top of the tunnel face. The information entered in step S2 is stored in the cloud database 30.

[0025] (Step S3) New analysis models that differ from existing analysis models are added to the cloud database 30 by the analysis provider's operator after prior consultation with the user, and are provided as options on the numerical analysis site, and are set as selectable analysis models in the aforementioned step S1.

[0026] (Step S4) The operator sets the conditions for creating the analysis results database, which is created through a parametric study.

[0027] (Step S5) When an instruction is given by the operator, database construction unit 502 of analysis device 50 constructs an analysis result database according to the following procedure. The database construction unit 502 sets analytical physical property values ​​(step S5-1). The database construction unit 502 executes a direct analysis, for example, by the finite difference method or the finite element method, using the new analytical model provided in step S3 (step S5-2). The database construction unit 502 adds the results of the forward analysis to the analysis result database (step S5-3). If the database creation conditions set in step S4 are met, the next step S5-5 is executed, and if not, the above-mentioned step S5-1 is executed again (step S5-4). Finally, the database construction unit 502 executes machine learning based on the analysis result database (step S5-5). The machine learning algorithm is, for example, Light Gradient Boosting Machine (LightGBM). The training data is data selected from the analysis result database according to the analysis model, such as displacement and stress.

[0028] (Step S6) The evaluation unit 504 evaluates whether the accuracy of the trained machine learning model meets predetermined evaluation criteria, and if the evaluation criteria are met, step S7 is executed, and if the criteria are not met, step S4 is executed.

[0029] (Step S7) Once the data stored in the cloud database 30 has been updated and all the data necessary for analysis is available, the analytical property value estimation unit 506 performs a reverse analysis using the trained machine learning model to estimate the analytical property values ​​of the tunnel (step S7-1). Next, the analytical physical property value estimation unit 506 performs a forward analysis using the obtained analytical physical property values ​​(step S7-2) to obtain analytical values ​​(displacement) at the position of interest (point A, which is the measurement point) (step S7-3). An example of the analysis obtained in step S7-3 is shown in Figure 5. In Figure 5, the horizontal axis represents the distance from the face and the vertical axis represents vertical displacement, and good results are shown, with the analytical values ​​and measured values ​​matching well. Finally, the analytical property value estimation unit 506 compares the calculated analytical value with the measurement data, and determines whether the degree of agreement between them satisfies a determination criterion (step S7-4). If the degree of match is below a predetermined level (if there is substantially no match), step S8 is executed, and if it exceeds the predetermined level (if there is substantially a match), step S9 is executed.

[0030] (Step S8) The database construction unit 502 adds the analysis results to the analysis result database, and then the above-mentioned machine learning (step 5-5) is executed. In this way, if the degree of match does not meet the judgment criteria in step S7-4, the database construction unit 502 adds the numerical analysis results to the analysis result database and re-trains the machine learning model, and by repeating this process, the accuracy of the machine learning model is improved.

[0031] (Step S9) The analysis results (estimated analytical physical property values, etc.) by the analytical physical property value estimation unit 506 are stored in the cloud database 30 and provided to the user via the numerical analysis site. The analysis results include, for example, displacement and stress in addition to the estimated analytical physical property values.

[0032] (Step S10) After the user checks the results returned from analysis device 50, a second analysis model is selected for analyzing the displacement (amount of subsidence) at positions P1 to P5 shown in Fig. 4. The second analysis model selected is a model corresponding to the position of the house. The model selected by the user is stored in the cloud database 30.

[0033] (Step S11) Once the data stored in the cloud database 30 has been updated and all the data necessary for analysis has been collected, the analysis unit 508 performs a forward analysis based on the second analysis model and analysis property values ​​stored in the cloud database 30, and determines the displacement (amount of subsidence) of positions P1 to P5.

[0034] (Step S12) The results obtained by the analysis device 50 (displacements of positions P1 to P5) are stored in the cloud database 30 and are provided to the user via the numerical analysis site. Thereafter, when the user confirms that the displacements at the positions P1 to P5 are within predetermined reference values, excavation can proceed ahead of the face.

[0035] In this way, with this tunnel analysis system 10, the number of processes involving operators of analysis companies that provide analysis services is significantly reduced compared to conventional systems, thereby shortening the time it takes to present analysis results to users.

[0036] Although the embodiments of the present invention have been described above, the present invention is not limited to the above-described embodiments, and all changes in conditions that do not depart from the gist of the present invention are within the scope of application of the present invention. [Explanation of symbols]

[0037] 10 Tunnel Analysis System 12 User terminals 20 Web server for providing services 30 Cloud Database 50 Analyzer 502 Database Construction Department 504 Evaluation Department 506 Analysis property value estimation unit 508 Analysis Department

Claims

1. a web server that provides a numerical analysis site for receiving data necessary for the numerical analysis of civil engineering structures from users; a cloud database in which data received at the numerical analysis site is stored; an analysis device that performs a numerical analysis of the civil engineering structure based on the data stored in the cloud database and stores the results in the cloud database, the analysis device includes a database construction unit that constructs a database of numerical analysis results; an evaluation unit that verifies that the accuracy of a machine learning model trained using the data in the analysis result database as training data satisfies a predetermined evaluation standard; an analytical property value estimation unit that performs an inverse analysis using the trained machine learning model that has been confirmed by the evaluation unit to satisfy the evaluation criteria, estimates analytical property values ​​of the civil engineering structure, performs a forward analysis using the analytical property values, and determines the degree of agreement between the obtained numerical analysis results and measurement data corresponding to the numerical analysis results; an analysis unit that performs a numerical analysis using the analytical physical property value estimated by the analytical physical property value estimation unit when the degree of coincidence satisfies a predetermined criterion, A civil engineering structure analysis system in which, if the degree of match does not satisfy the judgment criterion, the database construction unit adds the numerical analysis results to the analysis result database and re-trains the machine learning model.

2. 2. The civil engineering structure analysis system according to claim 1, A civil engineering structure analysis system in which the analysis device verifies that the data stored in the cloud database has been updated at a predetermined interval.

3. 3. The civil engineering structure analysis system according to claim 2, A civil engineering structure analysis system in which the learning algorithm of the machine learning model is LightGBM.

4. a web server that provides a numerical analysis site for receiving data necessary for the numerical analysis of civil engineering structures from users; a data storage unit in which data received at the numerical analysis site is stored; an analysis device that performs a numerical analysis of the civil engineering structure based on the data stored in the data storage unit and stores the results in the data storage unit, the analysis device includes a database construction unit that constructs a database of numerical analysis results; an evaluation unit that verifies that the accuracy of a machine learning model trained using the data in the analysis result database as training data satisfies a predetermined evaluation standard; an analytical property value estimation unit that performs an inverse analysis using the trained machine learning model that has been confirmed by the evaluation unit to satisfy the evaluation criteria, estimates analytical property values ​​of the civil engineering structure, performs a forward analysis using the analytical property values, and determines the degree of agreement between the obtained numerical analysis results and measurement data corresponding to the numerical analysis results; an analysis unit that performs a numerical analysis at a predetermined position using the analytical physical property value estimated by the analytical physical property value estimation unit when the degree of coincidence satisfies a predetermined criterion, A civil engineering structure analysis system in which, if the degree of match does not satisfy the judgment criterion, the database construction unit adds the numerical analysis results to the analysis result database and re-trains the machine learning model.

5. An analysis device connected to a web server that provides a numerical analysis site for receiving data required for the numerical analysis of civil engineering structures from users and a data storage unit that stores the data received at the numerical analysis site, the analysis device performing a numerical analysis of civil engineering structures based on the data stored in the data storage unit, and storing the results in the data storage unit, a database construction means for constructing a database of the results of the numerical analysis; evaluation means for confirming that the accuracy of the machine learning model trained using the data in the analysis result database as training data satisfies a predetermined evaluation standard; an analytical property value estimation means for performing an inverse analysis using the trained machine learning model that has been confirmed by the evaluation means to satisfy the evaluation criteria, estimating analytical property values ​​of the civil engineering structure, performing a forward analysis using the analytical property values, and determining the degree of agreement between the obtained numerical analysis results and measurement data corresponding to the numerical analysis results; when the degree of coincidence satisfies a predetermined criterion, the analysis means functions as an analysis means for performing a numerical analysis at a predetermined position using the analysis physical property value estimated by the analysis physical property value estimation means; A program in which, if the degree of match does not satisfy the judgment criterion, the database construction means adds the numerical analysis result to the analysis result database and re-trains the machine learning model.

Images