Building resilience characteristic model estimation system and building soundness determination system

Abstract

To estimate a restoring force characteristic model in which the present state of a building is more accurately expressed on the basis of time history data observed for the building.SOLUTION: A building restoring force characteristic model estimation system includes an observation data acquisition unit that acquires relative displacement time history data Dexp at the time of observation and load time history data Qexp at the time of observation, a parallel model construction unit that arranges a first restoring force characteristic model and a second restoring force characteristic model related to a building in parallel, sets a value to each of parameters, and constructs a parallel model 8, and inputs the relative displacement time history data Dexp at the time of observation to the parallel model 8. A response analysis unit that performs response analysis and acquires load-time history data Qana at the time of analysis, and a parallel model update unit that updates the parallel model 8 by searching for a value that reduces the objective function and updating each search parameter to the searched value are provided, and the estimated restoring force characteristic model is estimated by repeating the processes of the response analysis unit and the parallel model update unit.SELECTED DRAWING: Figure 3

Description

[Technical Field]

[0001] The present invention relates to a building restoring force characteristic model estimation system and a building health assessment system. [Background technology]

[0002] Various building health monitoring systems have been proposed that can determine the health of a building, such as the degree of damage to the building, after an earthquake occurs, without having to directly inspect the building. For example, Patent Document 1 discloses a configuration including a plurality of acceleration sensors installed on multiple floors of a building, a recording unit that calculates and analyzes the seismic intensity of each floor and the damage assessment of the building based on the detection data from the plurality of acceleration sensors within the building, records the analysis results, and a notification means that notifies the analysis results sent from the recording unit. Patent document 2 also discloses a configuration in which vibration sensors are installed at the joints of multiple structural members that form the structural frame of a structure, and detection information from vibration sensors installed at other joints where the opposite ends of each structural member joined to one end of each structural member joined at a joint is input, and the presence or absence of damage to the structural members and the extent of the damage are detected based on the input / output relationship of the dynamic characteristics, with the vibration sensors at the joints being used as output. Furthermore, Patent Document 3 discloses a building health assessment system that includes an earthquake detection unit that has a building bottom sensor that is installed under the building and detects acceleration, and an earthquake early warning receiver that receives earthquake early warnings, a plurality of sensors that are installed at multiple locations on the building and measure the impact of an earthquake on the building at each location, and a health estimation unit that estimates and evaluates the health of the building based on the measurement results, wherein the earthquake detection unit activates the plurality of sensors when the building bottom sensor detects initial tremors or the earthquake early warning receiver receives an earthquake early warning, and the plurality of sensors measure the impact of the earthquake on the building from before the main vibration arrives on the building until after it arrives.

[0003] The above-mentioned health monitoring system is intended to determine the damage caused to a building as a result of an actual earthquake. However, there is a need to predict and understand the damage that may occur to a building in the event of a future earthquake. To achieve this, a restoring force model that represents the building can be constructed based on various data from the design stage or on time history data observed during an actual earthquake. Then, seismic motions corresponding to the expected earthquake are input into this restoring force model to calculate the building's response, thereby predicting and evaluating the building's health. For example, in the case of a reinforced concrete structure, the well-known TAKEDA model can be used as the restoring force model. In this case, various parameters, such as initial stiffness, are adjusted so that the TAKEDA model approximates the restoring force characteristics of the building.

[0004] Here, the restoring force characteristics of a building change over time and due to the experience of disasters. Therefore, even if a restoring force characteristic model is constructed based on various data at the time of design, the above-mentioned changes are not reflected in the restoring force characteristic model, and the restoring force characteristic model does not represent the current state of the building. For this reason, even if the restoring force characteristic model constructed as described above is used to predict and evaluate the soundness of a building in the event of a future earthquake, there is a possibility that the accuracy will not be high. Therefore, in order to construct a restoring force characteristic model that represents the current state of the building, it is desirable to construct the restoring force characteristic model based on time history data observed for the building when an actual earthquake occurred.

[0005] However, the restoring force characteristics of buildings such as those described above can exhibit more complex behavior depending on the degree of change in the building, and it can be difficult to simply express the restoring force characteristics of a building using a single, well-known restoring force characteristic model. For example, in the case of a reinforced concrete building, repeated loading of large deformations can cause cracks in the concrete, resulting in large deformations when the load is reversed, resulting in a slip characteristic. In such cases, when using the TAKEDA model as the restoring force characteristic model as described above, it is not easy to reflect the slip characteristic simply by adjusting the parameters. There is a demand for a building restoring force characteristic model estimation system that estimates a restoring force characteristic model that more accurately represents the current state of a building based on time history data observed for the building, and a building health assessment system that estimates the health of a building using the restoring force characteristic model estimated by the restoring force characteristic model. [Prior art documents] [Patent documents]

[0006] [Patent Document 1] Japanese Patent Application Laid-Open No. 2013-254239 [Patent Document 2] Japanese Patent Application Laid-Open No. 2015-4526 [Patent Document 3] Japanese Patent Publication No. 2020-143895 Summary of the Invention [Problem to be solved by the invention]

[0007] The problem to be solved by the present invention is to provide a building restoring force characteristic model estimation system that estimates a restoring force characteristic model that more accurately represents the current state of a building based on time history data observed for the building, and a building health assessment system that estimates the health of a building using the restoring force characteristic model estimated by the restoring force characteristic model. [Means for solving the problem]

[0008] In order to solve the above-mentioned problems, the present invention employs the following means: That is, the present invention is a building restoring force characteristic model estimation system that estimates a restoring force characteristic model of a building based on time history data observed for the building, the system including an observation data acquisition unit that acquires relative displacement time history data at the time of observation and load time history data at the time of observation from the time history data, a parallel model construction unit that arranges a first restoring force characteristic model and a second restoring force characteristic model related to the building in parallel as springs, and sets values ​​for parameters that express each of the first restoring force characteristic model and the second restoring force characteristic model to construct a parallel model, and a parallel model construction unit that inputs the relative displacement time history data at the time of observation to the parallel model and constructs a parallel model. and a parallel model updating unit that searches for values ​​of search parameters to be searched for when estimating the restoring force characteristic model, among the parameters, that reduce an objective function expressed by the residual between the load time history data at the time of observation and the load time history data at the time of analysis, and updates each of the search parameters to the searched value, thereby updating the parallel model, and is characterized in that the restoring force characteristic model is estimated by repeating the processing of the response analysis unit and the parallel model updating unit. According to the above configuration, first, relative displacement time history data at the time of observation and load time history data at the time of observation are obtained from time history data observed for the building when an earthquake actually occurs. Next, the first restoring force characteristic model and the second restoring force characteristic model related to the target building are arranged in parallel as springs, and values ​​are set for the parameters that represent each of the first restoring force characteristic model and the second restoring force characteristic model to construct a parallel model. For example, if the building is made of reinforced concrete and slip properties need to be taken into consideration, it is conceivable to use a well-known restoring force characteristic model for reinforced concrete buildings, such as the TAKEDA model, as the first restoring force characteristic model, and a slip model that represents slip properties as the second restoring force characteristic model. The relative displacement time history data at the time of observation is input to such a parallel model, and response analysis is performed to obtain the load time history data at the time of analysis. Then, for each of the search parameters to be searched for when estimating the restoring force characteristic model among the parameters of the first restoring force characteristic model and the second restoring force characteristic model, a value is searched for that minimizes the objective function expressed by the residual between the load time history data at the time of observation and the load time history data at the time of analysis. Furthermore, the parallel model is updated by updating each of the search parameters to the searched value. In this way, the parallel model updated to minimize the objective function is such that when the relative displacement time history data at the time of observation observed in the target building is input into the parallel model and a response analysis is performed, the load time history data at the time of analysis obtained is closer to the load time history data at the time of observation based on the data actually observed in the building than the parallel model before the update. Therefore, the updated parallel model more accurately represents the current state of the target building than the parallel model before the update. As a result of repeating such response analysis and updating of the parallel model based on the results, the parallel model that is finally updated and generated is one in which the load time history data at the time of analysis, which is obtained when the relative displacement time history data observed for the building at the time of observation is input into the parallel model and a response analysis is performed, is sufficiently close to the load time history data at the time of observation that is based on the data actually observed for the building, and it is considered that the parallel model reflects in detail the actual and latest state of the building at the time the time history data was obtained. In this way, it is possible to provide a building restoring force characteristic model estimation system that estimates a restoring force characteristic model that more accurately represents the current state of the building.

[0009] In one aspect of the present invention, the first restoring force characteristic model represents the restoring force characteristics of a reinforced concrete building, and the search parameters of the first restoring force characteristic model include initial stiffness, and the second restoring force characteristic model represents the restoring force characteristics of a reinforced concrete building taking into account slip characteristics, and the search parameters of the second restoring force characteristic model include initial stiffness, yield displacement, and post-yield stiffness reduction rate, which is the post-yield stiffness divided by the initial stiffness of the second restoring force characteristic model. According to the above configuration, the first restoring force characteristic model represents the restoring force characteristics of a reinforced concrete building, and the second restoring force characteristic model represents the restoring force characteristics of a reinforced concrete building taking into account slip characteristics. This makes it possible to appropriately realize parallel models in cases where the building is a reinforced concrete building and it is desired to take into account slip characteristics.

[0010] In another aspect of the present invention, the observation data acquisition unit calculates the hysteretic absorbed energy at the time of observation from the relative displacement time history data at the time of observation and the load time history data at the time of observation, the response analysis unit calculates the hysteretic absorbed energy at the time of analysis from the relative displacement time history data at the time of observation and the load time history data at the time of analysis, and the objective function is expressed by the residual and the difference between the hysteretic absorbed energy at the time of observation and the hysteretic absorbed energy at the time of analysis. According to the above configuration, the objective function is expressed by the difference between the hysteretic absorbed energy at the time of observation and the hysteretic absorbed energy at the time of analysis, in addition to the residual between the load time history data at the time of observation and the load time history data at the time of analysis, which makes it possible to reduce the number of solutions to be searched.

[0011] In another aspect of the present invention, the parallel model construction unit generates a plurality of combinations of values ​​by selecting the values ​​one by one from each of the search parameters, and constructs the parallel model by setting the value included in each of the plurality of combinations to construct the parallel model, thereby constructing the plurality of parallel models; the response analysis unit performs a response analysis for each of the plurality of combinations using the parallel model constructed corresponding to the combination, acquires load-time history data at the time of the analysis, and calculates hysteretic absorbed energy at the time of the analysis; the parallel model update unit calculates the objective function for each of the plurality of combinations using the corresponding load-time history data at the time of the analysis and the hysteretic absorbed energy at the time of the analysis, and generates a plurality of new combinations of the values ​​of the search parameters based on the combination for which the calculated objective function is small; and updates the plurality of parallel models by setting the value included in each of the newly generated combinations to generate the parallel model. According to the above configuration, the parallel model construction unit first generates a plurality of combinations of values ​​by selecting one value from each of the search parameters, and constructs a parallel model by setting the values ​​included in each of the plurality of combinations. Furthermore, the response analysis unit performs a response analysis for each of the plurality of combinations using the parallel model constructed corresponding to the combination, acquires load time history data during the analysis, and calculates the hysteretic absorbed energy during the analysis. The parallel model update unit then calculates an objective function for each of the multiple combinations using the corresponding load-time history data during analysis and the historical absorbed energy during analysis. If there is a combination that optimizes the objective function, it is considered that the combination has a value close to the combination with the small objective function. Therefore, the parallel model update unit generates multiple new combinations of search parameter values ​​based on the combination with the smallest calculated objective function among the multiple combinations, and updates the multiple parallel models by generating parallel models by setting the values ​​included in each of the newly generated combinations. Thereafter, the response analysis unit performs a response analysis on each of the updated parallel models, obtains load time history data during the analysis for each of the multiple combinations, and calculates the historical absorbed energy during the analysis. In this way, by repeating the processing of the response analysis unit and the parallel model update unit, it is possible to efficiently find a combination of search parameter values ​​that appropriately reduces the objective function, and a parallel model for which the combination of search parameter values ​​is set.

[0012] The present invention also provides a building soundness assessment system comprising: a building restoring force characteristic model estimation system as described above; an earthquake response analysis unit that inputs time history data of an anticipated earthquake to the restoring force characteristic model estimated by the building restoring force characteristic model estimation system, performs a response analysis, and acquires the response of the building when the anticipated earthquake occurs; and a soundness assessment unit that estimates the amount of deformation of each story of the building based on the response, and assesses the soundness of the building when the anticipated earthquake occurs based on the amount of deformation. In the above configuration, by inputting time history data corresponding to an earthquake of a scale that is thought to occur in the future as a predicted earthquake into the restoring force characteristic model that represents the current state of the building estimated by the building restoring force characteristic model estimation system, it is possible to determine and predict the soundness of the current building if this predicted earthquake were to occur in the future. Furthermore, the restoring force characteristic model more accurately represents the current state of the building due to the building restoring force characteristic model estimation system described above. Therefore, if a predicted earthquake occurs in the future, the state of the building and its soundness can be more accurately assessed and predicted. [Effects of the Invention]

[0013] According to the present invention, it is possible to provide a building restoring force characteristic model estimation system that estimates a restoring force characteristic model that more accurately represents the current state of a building based on time history data observed for the building, and a building health assessment system that estimates the health of a building using the restoring force characteristic model estimated by the restoring force characteristic model. [Brief explanation of the drawings]

[0014] [Figure 1] FIG. 1 is a diagram illustrating an example of a building for which a restoring force characteristic model is to be estimated in a building restoring force characteristic model estimation system according to an embodiment of the present invention. [Figure 2] FIG. 2 is a block diagram of the building restoring force characteristic model estimation system. [Figure 3] FIG. 2 is a diagram showing a data flow in the building restoring force characteristic model estimation system. [Figure 4] FIG. 10 is a diagram showing an example of relative displacement time history data during observation. [Figure 5] FIG. 2 is an explanatory diagram of a parallel model constructed in the building restoring force characteristic model estimation system. [Figure 6] FIG. 2 is an explanatory diagram of a first restoring force characteristic model constituting a parallel model. [Figure 7] FIG. 10 is an explanatory diagram of a second restoring force characteristic model constituting a parallel model. [Figure 8] 1 is a block diagram of a building health assessment system using a building restoring force characteristic model estimation system according to an embodiment of the present invention. FIG. [Figure 9] 10 is a flowchart of a restoring force characteristic model estimation method in the building restoring force characteristic model estimation system of the embodiment. [Figure 10] 10 is a flowchart of a building soundness determination method in the building soundness determination system of the embodiment. [Figure 11] FIG. 10 is a diagram showing load time history data observed using a scaled-down test specimen in a vibration experiment, and load time history data obtained by inputting the same time history data as in the vibration experiment into a restoring force characteristic model estimated by the building restoring force characteristic model estimation system of the above embodiment. [Figure 12] FIG. 10 is a diagram showing the hysteretic absorbed energy observed by a scaled-down test specimen in a vibration experiment, and the hysteretic absorbed energy obtained by inputting the same time history data as in the vibration experiment into the restoring force characteristic model estimated by the building restoring force characteristic model estimation system of the above embodiment. [Figure 13] FIG. 10 is a diagram showing the relationship between load and deformation observed using a scaled-down test specimen in a vibration experiment, and the relationship between load and deformation obtained by inputting the same time history data as in the vibration experiment into the restoring force characteristic model estimated by the building restoring force characteristic model estimation system of the above embodiment. [Figure 14] FIG. 10 is a diagram showing the relationship between load and deformation observed using a scaled-down test specimen in a vibration experiment for the after-sway section, and the relationship between load and deformation obtained by inputting the same time history data as in the vibration experiment into the restoring force characteristic model estimated by the building restoring force characteristic model estimation system of the above embodiment. [Figure 15] This figure shows the relationship between load and deformation on each floor when an estimated restoring force characteristic model was set for each floor of the mass system model and a simulation analysis of earthquake response was performed. [Figure 16]FIG. 10 is a diagram showing the maximum inter-story deformation for each of the restoring force characteristic models inferred for each story. [Figure 17] FIG. 10 is a diagram showing the maximum story shear force for each of the restoring force characteristic models estimated for each story. DETAILED DESCRIPTION OF THE INVENTION

[0015] Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. FIG. 1 is a diagram showing an example of a building for which a restoring force characteristic model is to be estimated in a building restoring force characteristic model estimation system according to an embodiment of the present invention. 1, a building 100 has 20 floors, for example, from the first floor (1F) to the twentieth floor (20F). An acceleration sensor 2 is provided on the floor of each floor. In this embodiment, the building 100 is made of reinforced concrete. The restoring force characteristics of the building 100 change over time and due to disaster experience. Depending on the degree of change, the restoring force characteristics of the building may exhibit more complex properties, and therefore it may be difficult to simply express the restoring force characteristics of the building 100 using a single well-known restoring force characteristic model. In the case of a reinforced concrete building 100 as in this embodiment, repeated loading of large deformations may cause cracks in the concrete, resulting in slip characteristics in which large deformation occurs when the load is reversed. The building restoring force characteristic model estimation system of this embodiment constructs a restoring force characteristic model that accurately represents the current state of the building 100 even in the above-mentioned cases.

[0016] FIG. 2 is a block diagram of the building restoring force characteristic model estimation system. The building restoring force characteristic model estimation system 1 includes a plurality of acceleration sensors 2 as described above, which are provided on each floor, and a computing device 3. When an earthquake occurs, each acceleration sensor 2 observes time history data for the floor on which the acceleration sensor 2 is installed. In this embodiment, the time history data is acceleration time history data. The building restoring force characteristic model estimation system 1 estimates a restoring force characteristic model of the building 100 based on the time history data observed by each acceleration sensor 2 when an earthquake occurs. The computing device 3 is a computer terminal such as a server or a personal computer, and performs required functions by executing a preset program. Functionally, the computing device 3 includes an observation data acquisition unit 4, a parallel model construction unit 5, a response analysis unit 6, and a parallel model update unit 7.

[0017] Fig. 3 is a diagram showing the flow of data in the building restoring force characteristic model estimation system Fig. 4 is a diagram showing an example of relative displacement time history data at the time of observation. The observation data acquisition unit 4 receives the time history data of each floor observed by the acceleration sensor 2 on each floor as the acceleration time history data Aexp at the time of observation for each floor. The observation data acquisition unit 4 acquires the displacement time history data for each floor by double-integrating the acceleration time history data Aexp at the time of observation for each floor. The observation data acquisition unit 4 acquires the relative displacement time history data Dexp at the time of observation for each floor by subtracting the displacement time history data of each floor from the displacement time history data of the floor located one level above. For example, for the 8th floor (8F), the observation data acquisition unit 4 calculates the relative displacement time history data Dexp at the time of observation for the 8th floor (8F) by subtracting the displacement time history data obtained by double-integrating the acceleration time history data Aexp at the time of observation observed by the acceleration sensor 2 on the 9th floor (9F) from the displacement time history data obtained by double-integrating the acceleration time history data Aexp at the time of observation observed by the acceleration sensor 2 on the 9th floor (9F). In this manner, in this embodiment, the relative displacement time history data Dexp at the time of observation of each story is the inter-story deformation time history data of that story.

[0018] Next, the observation data acquisition unit 4 acquires the load time history data Qexp at the time of observation for each story. For the top floor, the observation data acquisition unit 4 calculates and acquires the load time history data Qexp at the time of observation of the top floor by multiplying the mass of that floor by the acceleration time history data Aexp at the time of observation of that floor. For floors below the top floor, the observation data acquisition unit 4 calculates and acquires the load time history data Qexp at the time of observation for that floor by adding the time history data obtained by multiplying the acceleration time history data Aexp at the time of observation for that floor by the mass of that floor, and the load time history data Qexp at the time of observation acquired for the floor one level above. In the above, the mass of each story is calculated by dividing the weight of each story by the acceleration due to gravity. In this way, the observation data acquisition unit 4 calculates and acquires the relative displacement time history data Dexp and the load time history data Qexp at the time of observation from the time history data of each floor obtained by the acceleration sensor 2 (acceleration time history data Aexp at the time of observation).

[0019] Furthermore, the observation data acquisition unit 4 calculates the hysteretic absorbed energy Eexp at the time of observation for each floor as the area of ​​the hysteretic loop using the relative displacement time history data Dexp at the time of observation and the load time history data Qexp at the time of observation, using the following equation (1).

number

[0020] As will be explained later, the parameters of the restoring force characteristic model are searched for to minimize both the difference ΔE between the hysteretic absorbed energy Eexp during observation and the hysteretic absorbed energy Eana (described later) during analysis, and the residual ΔQ between the load time history data Qexp during observation and the load time history data Qana (described later) during analysis. In this search, the relative displacement time history data Dexp during observation (shown in Figure 4) shows that slip characteristics are prominent in the restoring force characteristics at locations where the relative displacement is large. Therefore, in order to estimate a model that can accurately represent the slip characteristics, the search targets sections with large relative displacement. Specifically, a range of five cycles before and after the maximum value of the relative displacement is extracted from the relative displacement time history data Dexp during observation, and the above equation (1) and subsequent processing are calculated using only the data included in this range. As described above, in this embodiment, the observation data acquisition unit 4 calculates the above formula (1) for a section R (see FIG. 4 ) in which the deformation (inter-story displacement) is larger than other cycles, for example, about five cycles, in the relative displacement time history data Dexp during observation. More specifically, the time when the amplitude value becomes 0, for example, five cycles before the time Tm when the amplitude becomes maximum, is set as the start time T1. Furthermore, the time when the amplitude value becomes 0, for example, nine cycles after time T1 (the tenth cycle when start time T1 is the first cycle), is set as the end time T2. The period from start time T1 to end time T2 set in this way is set as section R, and the observation data acquisition unit 4 calculates the above formula (1) for this section R. In this way, the observation data acquisition unit 4 calculates the hysteretic absorbed energy Eexp at the time of observation in section R from the portions within section R of the relative displacement time history data Dexp at the time of observation and the load time history data Qexp at the time of observation.

[0021] The parallel model construction unit 5 constructs a parallel model as an initial state of the restoring force characteristic model estimated by the building restoring force characteristic model estimation system 1. In this embodiment, for each of all floors of the building 100, the parallel model construction unit 5 constructs a parallel model 8 corresponding to that floor. FIG. 5 is an explanatory diagram of a parallel model constructed in the building restoring force characteristic model estimation system. As already explained, it may be difficult to express the restoring force characteristics of the building 100 using one well-known restoring force characteristic model. For this reason, in this embodiment, the restoring force characteristic model corresponding to each story of the building 100 is expressed by combining a first restoring force characteristic model 9A and a second restoring force characteristic model 9B, both of which are related to the building 100. Specifically, the parallel model construction unit 5 expresses the restoring force characteristic model corresponding to each story of the building 100 as a parallel model 8 in which the first restoring force characteristic model 9A and the second restoring force characteristic model 9B are arranged in parallel as springs.

[0022] In this embodiment, the building 100 is made of reinforced concrete. Therefore, in this embodiment, the first restoring force characteristic model 9A represents the restoring force characteristics of a reinforced concrete building. More specifically, in this embodiment, the TAKEDA model, which is well known as a restoring force characteristic model for reinforced concrete buildings, is used as the first restoring force characteristic model 9A. FIG. 6 is an explanatory diagram of the first restoring force characteristic model constituting the parallel model. The first restoring force characteristic model 9A, which is a TAKEDA model, is a trilinear type, with three lines LA1, LA2, and LA3 continuing with changing slopes through the first yield point PA1 and second yield point PA2, taking cracks and yielding into consideration. As already explained, a reinforced concrete building will exhibit slip characteristics when subjected to repeated loading of large deformations. For this reason, in this embodiment, a slip model that expresses the restoring force characteristics of a reinforced concrete building in consideration of slip characteristics is used as the second restoring force characteristic model 9B. FIG. 7 is an explanatory diagram of the second restoring force characteristic model constituting the parallel model. The second restoring force characteristic model 9B, which is a slip model, is a bilinear type in which two lines LB1 and LB2 are continuous with changing slopes through one yield point PB1.

[0023] Next, the parallel model constructing unit 5 sets values ​​(initial values) for each of the parameters that represent the first restoring force characteristic model 9A. 6, the parameters representing the first restoring force characteristic model 9A include the initial stiffness (k1_t), which is the slope of the line LA1 extending from the origin, the first yield displacement (d1_t), which is the displacement at the first yield point PA1, the second yield displacement (d2_t), which is the displacement at the second yield point PA2, the stiffness reduction rate after the first yield (k2_t / k1_t), which is the value obtained by dividing the stiffness after yielding at the first yield point PA1, i.e., the slope of line LA2 (k2_t), by the initial stiffness, i.e., the slope of line LA1, the stiffness reduction rate after the second yield (k3_t / k1_t), which is the value obtained by dividing the stiffness after yielding at the second yield point PA2, i.e., the slope of line LA3 (k3_t), by the initial stiffness, i.e., the slope of line LA1, a parameter α that determines the unloading stiffness, and the stiffness reduction rate β during the inner loop iteration. The parallel model construction unit 5 sets values ​​for each of these parameters. As will be described later, the building restoring force characteristic model estimation system 1 searches for parameters that express the first restoring force characteristic model 9A (and the second restoring force characteristic model 9B), thereby estimating a restoring force characteristic model that appropriately expresses the building 100. In this embodiment, the search parameter of the first restoring force characteristic model 9A that is the target of this search is the initial stiffness (k1_t). In this way, the search parameters of the first restoring force characteristic model 9A include the initial stiffness.

[0024] For the initial stiffness (k1_t), which is a search parameter of the first restoring force characteristic model 9A, the search range can be set as a range between the lower limit and the upper limit, with the lower limit being the value obtained by multiplying the design model value or the average value of values ​​previously obtained for buildings similar to the target building 100 by 0.2 and the upper limit being the value obtained by multiplying the design model value or the average value of values ​​previously obtained for buildings similar to the target building 100 by 1.5. The parallel model construction unit 5 selects, for example, an arbitrary value from this search range and sets it as the value (initial value) of the initial stiffness (k1_t) in the parallel model 8 of each hierarchy. In this embodiment, for the parameters of the first restoring force characteristic model 9A that are not search parameters, namely the first yield displacement (d1_t), the second yield displacement (d2_t), the stiffness reduction rate after the first yield (k2_t / k1_t), and the stiffness reduction rate after the second yield (k3_t / k1_t), the values ​​of the design model or the average values ​​obtained in the past for buildings similar to the target building 100 are used. Furthermore, for the parameter α that determines the unloading stiffness and the stiffness reduction rate β during the inner loop iteration, the standard values ​​α = -0.4 and β = 0.7 can be used, respectively. For each of the above parameters that are not search parameters, the parallel model construction unit 5 sets the above values ​​in the parallel model 8 of each layer.

[0025] Similarly, the parallel model constructing unit 5 sets values ​​(initial values) for each of the parameters that represent the second restoring force characteristic model 9B. 7, the parameters expressing the second restoring force characteristic model 9B include the initial stiffness (k1_s), which is the slope of the line LB1 extending from the origin, the yield displacement (d1_s), which is the displacement at the yield point PB1, and the stiffness reduction rate after yield (k2_s / k1_s), which is the value obtained by dividing the stiffness after yielding at the yield point PB1, i.e., the slope (k2_s) of the line LB2, by the initial stiffness, i.e., the slope of the line LB1. The parallel model construction unit 5 sets values ​​for each of these. As will be described later, the building restoring force characteristic model estimation system 1 searches for parameters that express the second restoring force characteristic model 9B (and the first restoring force characteristic model 9A), thereby estimating a restoring force characteristic model that appropriately expresses the building 100. In this embodiment, the search parameters of the second restoring force characteristic model 9B that are the subject of this search are the initial stiffness (k1_s), the yield displacement (d1_s), and the stiffness reduction rate (k2_s / k1_s). Thus, the search parameters of the second restoring force characteristic model 9B include the initial stiffness, the yield displacement, and the stiffness reduction rate after yield, which is the stiffness after yield divided by the initial stiffness of the second restoring force characteristic model 9B.

[0026] For the initial stiffness (k1_s), which is a search parameter of the second restoring force characteristic model 9B, the search range can be set as a range between the lower limit and the upper limit, with the value set as the initial stiffness (k1_t) of the first restoring force characteristic model 9A multiplied by 0 as the lower limit and the value set as the initial stiffness (k1_t) of the first restoring force characteristic model 9A multiplied by 1.5 as the upper limit. The parallel model construction unit 5 selects, for example, an arbitrary value from this search range and sets it as the value (initial value) of the initial stiffness (k1_s) in the parallel model 8 of each hierarchical level. The search range for the yield displacement (d1_s), which is a search parameter for the second restoring force characteristic model 9B, can be set to a range between the lower limit of 0.1 and the upper limit of 0.5. The parallel model construction unit 5 selects, for example, an arbitrary value from this search range and sets it as the value (initial value) of the yield displacement (d1_s) for the parallel model 8 of each layer. With regard to the stiffness reduction rate (k2_s / k1_s), which is a search parameter of the second restoring force characteristic model 9B, the search range can be set as a range between the lower limit and the upper limit, with the lower limit being a value obtained by multiplying the story height of the story corresponding to the parallel model 8 constructed using the second restoring force characteristic model 9B by 0.005 and the upper limit being a value obtained by multiplying the story height by 0.02. The parallel model construction unit 5 selects, for example, an arbitrary value from this search range and sets it as the value (initial value) of the stiffness reduction rate (k2_s / k1_s) in the parallel model 8 of each story.

[0027] In the manner described above, a parallel model 8 is constructed for each stratum. The response analysis unit 6 inputs the relative displacement time history data Dexp at the time of observation acquired for that stratum to each of the parallel models 8 constructed for each stratum. More specifically, the response analysis unit 6 fixes one end 8a of each of the parallel models 8 constructed for each stratum, and inputs the relative displacement time history data Dexp at the time of observation to the other end 8b. In this manner, the response analysis unit 6 performs response analysis on each of the parallel models 8 constructed for each stratum. The response analysis outputs and obtains the load time history data Qana during the analysis. Furthermore, the response analysis unit 6 calculates the hysteretic absorbed energy Eana at the time of analysis for each story using the relative displacement time history data Dexp at the time of observation and the load time history data Qana at the time of analysis as the area of ​​the hysteretic loop using the following equation (2).

number

[0028] The parallel model update unit 7 expresses the objective function for each layer using the residual ΔQ between the load time history data Qexp at the time of observation and the load time history data Qana at the time of analysis, and the difference ΔE between the hysteretic absorbed energy Eexp at the time of observation and the hysteretic absorbed energy Eana at the time of analysis. More specifically, the parallel model update unit 7 formulates the residual ΔQ between the load time history data Qexp at the time of observation and the load time history data Qana at the time of analysis as in the following equation (3).

number

[0029] Furthermore, the parallel model update unit 7 expresses the difference ΔE between the hysteresis energy absorption Eexp during observation and the hysteresis energy absorption Eana during analysis using the following equation (4).

number

[0030] In this embodiment, the parallel model update unit 7 uses both the residual ΔQ and the difference ΔE to express the objective function f as the following equation (5).

number

[0031] The parallel model update unit 7 formulates the objective function f for each story as shown in equation (5). Here, if the initial stiffness (k1_t), which is a search parameter of the first restoring force characteristic model 9A of the parallel model 8, and the initial stiffness (k1_s), yield displacement (d1_s), and stiffness reduction rate (k2_s / k1_s), which are search parameters of the second restoring force characteristic model 9B, are close to the actual state of the target building 100 at the time of observing the acceleration time history data Aexp, the load time history data Qana at the time of analysis output as the result of the response analysis by the response analysis unit 6 should be close to the load time history data Qexp at the time of observation, and therefore the residual ΔQ should be small. Furthermore, if the load time history data Qana at the time of analysis and the load time history data Qexp at the time of observation are close, the hysteretic absorbed energy Eexp at the time of observation and the hysteretic absorbed energy Eana at the time of analysis, which are calculated based on these data, should also be close, and the difference ΔE should also be small. Therefore, the closer the values ​​of the above search parameters are to the current state of the target building 100, the smaller the value of the objective function f will be. Based on this idea, the parallel model update unit 7 searches for values ​​of each search parameter that will reduce the objective function f.

[0032] As already explained, the parallel model 8 inputs the relative displacement time history data Dexp at the time of observation, performs a response analysis, and outputs the load time history data Qana at the time of analysis. For this reason, it is also possible to formulate the objective function f using only a term related to the residual ΔQ between the load time history data Qexp at the time of observation, which is calculated based on the relative displacement time history data Dexp at the time of observation, and the load time history data Qana at the time of analysis, for the purpose of comparing these. In other words, the objective function f may not include a term related to the difference ΔE between the hysteretic absorbed energy Eexp at the time of observation and the hysteretic absorbed energy Eana at the time of analysis, and may be expressed using only the residual ΔQ between the load time history data Qexp at the time of observation and the load time history data Qana at the time of analysis. However, in this case, there is a possibility that many combinations of search parameter values ​​that optimize the objective function f will be searched for. In order to prevent a large number of search combinations of search parameter values ​​from being searched for, in this embodiment, the objective function f includes both the residual ΔQ and the difference ΔE as shown in the above formula (5). This narrows down the situations in which the objective function f is optimal, and the number of search parameter combinations that optimize the objective function f converges to a small number.

[0033] The parallel model update unit 7 searches for values ​​of the search parameters for each layer such that the objective function f is small, for example, optimal. The parallel model update unit 7 updates the values ​​of the search parameters of the parallel model 8 for each layer to the values ​​searched for as described above, thereby updating the parallel model 8 for that layer. When the parallel model 8 is updated, the response analysis unit 6 performs response analysis again on the updated parallel model 8, and the parallel model update unit 7 evaluates the objective function f based on the result of the response analysis. In this way, the processes of response analysis in the response analysis unit 6 and evaluation of the objective function f in the parallel model update unit 7 are repeatedly executed for each layer. When evaluating the objective function f for each floor, the parallel model update unit 7 determines whether the evaluated value is equal to or less than a predetermined threshold. If the objective function f is equal to or less than the threshold, the parallel model update unit 7 determines that the parallel model 8 represents the current state of the target building 100 with sufficient accuracy, and outputs the parallel model 8 as the restoring force characteristic model estimated for the target building 100. 9, the processing of the response analysis unit 6 and the parallel model update unit 7 is repeated until the objective function f expressed by the residual ΔQ between the load time history data Qexp at the time of observation and the load time history data Qana at the time of analysis becomes equal to or less than the threshold value. However, this is not limited to this. For example, when the repeated calculation is performed 1000 times, the objective function f becomes sufficiently smaller than the assumed threshold value. For this reason, it is also possible to terminate the repetition when the repeated processing has been performed a pre-specified number of times without determining whether the objective function f is equal to or less than the threshold value.

[0034] The above-described series of repetitive processes, including setting the values ​​of multiple (four in this embodiment) search parameters for the parallel model 8, performing a response analysis on the parallel model 8, searching for parameter values ​​that reduce the objective function f based on the execution results, and resetting (updating) the parallel model 8 using the searched values, can be solved by using a commercially available program or system equipped with an optimization algorithm, such as that provided by IDEJ Inc. under the product name: modeFRONTIER. In the following, a process for searching for parameters to be set in the parallel model 8 by applying, for example, an evolutionary computation method will be described as an example.

[0035] In this case, the parallel model construction unit 5 first generates a plurality of combinations of values ​​by selecting one value from each of the search parameters. Specifically, the parallel model construction unit 5 determines values ​​for each of the plurality of (e.g., four) search parameters, for example, randomly from the range of values ​​already described, to determine the combination of values, and repeats this process multiple times to generate a plurality of combinations of values. Then, the parallel model construction unit 5 sets a value included in each of these plurality of combinations to construct a parallel model 8 corresponding to the combination. The parallel model construction unit 5 executes the above-described process for each of the plurality of hierarchies. That is, the parallel model construction unit 5 constructs a parallel model 8 corresponding to each of the plurality of hierarchies.

[0036] Next, the response analysis unit 6 performs a response analysis for each of the plurality of combinations using the parallel model 8 constructed corresponding to the combination, and acquires load time history data Qana during analysis for each of the plurality of combinations. Furthermore, the response analysis unit 6 calculates, for each of the plurality of combinations, the hysteretic absorbed energy Eana during analysis using equation (2) based on the load time history data Qana during analysis acquired corresponding to that combination.

[0037] The parallel model update unit 7 calculates the objective function f for each of the plurality of combinations using the corresponding load time history data Qana at the time of analysis and the historical absorbed energy Eana at the time of analysis. For example, when a genetic algorithm is applied as an evolutionary computation method, the parallel model update unit 7 searches for the values ​​of each of the search parameters by applying processes such as selection, crossover, and mutation to multiple combinations. For example, as a selection process, the parallel model update unit 7 selects and acquires, for example, a predetermined number of combinations that provide small calculation results of the objective function f from among a plurality of combinations. Based on the multiple combinations selected as described above, the parallel model update unit 7 generates multiple new combinations of search parameter values ​​that are expected to reduce the objective function f, for example, by processing such as crossover or mutation.

[0038] The parallel model update unit 7 updates the parallel models 8 by generating parallel models 8 by setting the values ​​included in each of the newly generated combinations. By repeating this process, it is possible to search for values ​​of the search parameters that make the objective function f small, for example, optimal.

[0039] Next, a building soundness determination system using the building restoring force characteristic model estimation system 1 as described above will be described. The building health assessment system inputs time history data, such as data corresponding to an earthquake anticipated in the future, into the restoring force characteristic model estimated by the building restoring force characteristic model estimation system 1, and thereby assesses (predicts) the health of the building 100 when the anticipated earthquake occurs. FIG. 8 is a block diagram of a building soundness determination system using a building restoring force characteristic model estimation system according to an embodiment. The building health assessment system 10 is comprised of a computer terminal such as a server, a personal computer, or a tablet terminal, and performs required functions by executing a preset program. In addition to the building restoring force characteristic model estimation system 1 described above, the building health assessment system 10 functionally comprises an earthquake response analysis unit 11 and a health assessment unit 12.

[0040] The earthquake response analysis unit 11 inputs time history data of an expected earthquake into the restoring force characteristic model estimated by the building restoring force characteristic model estimation system 1, performs a response analysis, and acquires the response of the building 100 when the expected earthquake occurs. More specifically, the earthquake response analysis unit 11 inputs relative displacement time history data of the relevant story corresponding to the expected earthquake into each of the restoring force characteristic models of each story estimated by the building restoring force characteristic model estimation system 1, for example, to perform a response analysis, and acquires the response of each story when the expected earthquake occurs. The soundness determination unit 12 determines the soundness of the building 100 based on the response of the building 100 acquired by the earthquake response analysis unit 11. For example, the soundness determination unit 12 acquires relative displacement time history data for each story based on the response acquired for that story, and estimates the amount of deformation of that story. For each story, the soundness determination unit 12 compares, for example, the amount of deformation with a predetermined threshold, and if the amount of deformation is greater than the threshold, determines that there is a problem with the soundness of the building 100 if an anticipated earthquake occurs. The soundness assessment unit 12 may be configured to estimate the amount of deformation for each floor, calculate the inter-story deformation angle for that floor, compare this inter-story deformation angle with a predetermined threshold, and if the inter-story deformation angle is greater than the threshold, determine that there is a problem with the soundness of the building 100 if an anticipated earthquake occurs.

[0041] Next, a building restoring force characteristic model estimating method using the above-described building restoring force characteristic model estimating system 1 will be described with reference to FIGS. 1 to 8 and 9. FIG. FIG. 9 is a flowchart of a restoring force characteristic model estimation method in the building restoring force characteristic model estimation system of this embodiment. When an earthquake occurs, each acceleration sensor 2 observes time history data for the story on which the acceleration sensor 2 is installed. The observation data acquisition unit 4 receives the time history data of each floor observed by the acceleration sensor 2 of each floor as acceleration time history data Aexp at the time of observation for each floor (step S1). The observation data acquisition unit 4 calculates and acquires the relative displacement time history data Dexp and the load time history data Qexp at the time of observation from the time history data of each floor (acceleration time history data Aexp at the time of observation) obtained by the acceleration sensor 2 (step S2). Furthermore, the observation data acquisition unit 4 calculates the historical absorbed energy Eexp at the time of observation for each floor using equation (1).

[0042] The parallel model construction unit 5 constructs a parallel model as an initial state of the restoring force characteristic model estimated by the building restoring force characteristic model estimation system 1. In this embodiment, for each of all floors of the building 100, the parallel model construction unit 5 constructs a parallel model 8 corresponding to that floor (step S3). The response analysis unit 6 inputs the relative displacement time history data Dexp at the time of observation acquired for each story into each of the parallel models 8 constructed for each story. In this way, the response analysis unit 6 performs response analysis for each of the parallel models 8 constructed for each story (step S4). The response analysis outputs and obtains the load time history data Qana during the analysis. Furthermore, the response analysis unit 6 calculates the hysteretic absorbed energy Eana at the time of analysis for each story using the relative displacement time history data Dexp at the time of observation and the load time history data Qana at the time of analysis according to equation (2).

[0043] The parallel model update unit 7 expresses the objective function for each layer using the residual ΔQ between the load time history data Qexp at the time of observation and the load time history data Qana at the time of analysis, and the difference ΔE between the hysteretic absorbed energy Eexp at the time of observation and the hysteretic absorbed energy Eana at the time of analysis. When evaluating the objective function f for each layer, the parallel model update unit 7 determines whether the value of the evaluation result is equal to or less than a predetermined threshold value (step S5). If the objective function f is not equal to or less than the threshold for a certain layer (No in step S5), the parallel model update unit 7 searches for values ​​of each search parameter that make the objective function f small, for example, optimal. The parallel model update unit 7 updates the parallel model 8 for that layer by updating each value of the search parameters of the parallel model 8 to the searched values ​​(step S6). Thereafter, the process transitions to step S4, and thereafter, the response analysis in the response analysis unit 6 and the evaluation of the objective function f in the parallel model update unit 7 are repeatedly executed until the objective function f becomes equal to or less than the threshold in step S5. If the objective function f for a certain floor is equal to or less than the threshold value (Yes in step S5), the parallel model update unit 7 determines that the parallel model 8 represents the current state of the target building 100 with sufficient accuracy, and outputs the parallel model 8 as an estimated restoring force characteristic model for the target building 100 (step S7).

[0044] Next, a building soundness assessment method using the above-described building soundness assessment system 10 will be described with reference to FIG. FIG. 10 is a flowchart of a building soundness assessment method in the building soundness assessment system of this embodiment. First, the building restoring force characteristic model estimation system 1 estimates the restoring force characteristic model of the target building 100 (step S11). The earthquake response analysis unit 11 inputs the time history data of an expected earthquake into the restoring force characteristic model estimated by the building restoring force characteristic model estimation system 1, performs a response analysis, and obtains the response of the building 100 when the expected earthquake occurs (step S12). The soundness determining unit 12 determines the soundness of the building 100 based on the response of the building 100 acquired by the earthquake response analyzing unit 11 (step S13).

[0045] The building restoring force characteristic model estimation system 1 as described above is a building restoring force characteristic model estimation system 1 that estimates a restoring force characteristic model of the building 100 based on time history data (acceleration time history data Aexp) observed for the building 100, and includes an observation data acquisition unit 4 that acquires relative displacement time history data Dexp and load time history data Qexp at the time of observation from the time history data (acceleration time history data Aexp), and a parallel model 8 that arranges a first restoring force characteristic model 9A and a second restoring force characteristic model 9B related to the building 100 in parallel as springs, and sets values ​​for parameters that represent each of the first restoring force characteristic model 9A and the second restoring force characteristic model 9B. The system includes a parallel model construction unit 5 that constructs a parallel model to be used for estimating a restoring force characteristic model, a response analysis unit 6 that inputs the relative displacement time history data Dexp at the time of observation into the parallel model 8 and performs a response analysis to obtain the load time history data Qana at the time of analysis, and a parallel model update unit 7 that searches for values ​​that minimize an objective function f expressed by the residual ΔQ between the load time history data Qexp at the time of observation and the load time history data Qana at the time of analysis for each of the search parameters that are the search targets when estimating the restoring force characteristic model, and updates each of the search parameters to the searched values, thereby updating the parallel model 8. The restoring force characteristic model is estimated by repeating the processing of the response analysis unit 6 and the parallel model update unit 7. According to the above configuration, first, the relative displacement time history data Dexp and the load time history data Qexp at the time of observation are obtained from the time history data (acceleration time history data Aexp) observed for the building 100 when an earthquake actually occurs. Next, a first restoring force characteristic model 9A and a second restoring force characteristic model 9B related to the target building 100 are arranged in parallel as springs, and values ​​are set for the parameters that represent the first restoring force characteristic model 9A and the second restoring force characteristic model 9B, to construct a parallel model 8. For example, if the building 100 is made of reinforced concrete and slip properties need to be taken into consideration, it is conceivable to use a well-known restoring force characteristic model for reinforced concrete buildings 100, such as the TAKEDA model, as the first restoring force characteristic model 9A, and a slip model that represents slip properties as the second restoring force characteristic model 9B. Relative displacement time history data Dexp at the time of observation is input to such a parallel model 8, and response analysis is performed to obtain load time history data Qana at the time of analysis. Then, for each of the search parameters to be searched for when estimating the restoring force characteristic models among the parameters of the first restoring force characteristic model 9A and the second restoring force characteristic model 9B, a value is searched for that minimizes the objective function f, which is expressed by the residual ΔQ between the load time history data Qexp at the time of observation and the load time history data Qana at the time of analysis. Furthermore, the parallel model 8 is updated by updating each of the search parameters to the searched value. In this way, the parallel model 8 updated to minimize the objective function f is such that the load time history data Qana at the time of analysis, obtained when the relative displacement time history data Dexp at the time of observation observed in the target building 100 is input to the parallel model 8 and a response analysis is performed, is closer to the load time history data Qexp at the time of observation based on the data actually observed in the building 100 than the parallel model 8 before the update. Therefore, the updated parallel model 8 more accurately represents the current state of the target building 100 than the parallel model 8 before the update. As a result of repeating such response analysis and updating of the parallel model 8 based on the results, the parallel model 8 that is finally updated and generated is such that the load time history data Qana at the time of analysis obtained when the relative displacement time history data Dexp observed for the building 100 at the time of observation is input into the parallel model 8 and a response analysis is performed is sufficiently close to the load time history data Qexp at the time of observation based on the data actually observed for the building 100, and it is considered that the parallel model 8 reflects in detail the actual and latest state of the building 100 at the time the time history data (acceleration time history data Aexp) was obtained. In this way, it is possible to provide a building restoring force characteristic model estimation system 1 that estimates a restoring force characteristic model that more accurately represents the current state of the building 100.

[0046] Furthermore, the first restoring force characteristic model 9A represents the restoring force characteristics of the reinforced concrete building 100, and the search parameters of the first restoring force characteristic model 9A include initial stiffness, and the second restoring force characteristic model 9B represents the restoring force characteristics of the reinforced concrete building 100 taking into account slip characteristics, and the search parameters of the second restoring force characteristic model 9B include initial stiffness, yield displacement, and post-yield stiffness reduction rate, which is the post-yield stiffness divided by the initial stiffness of the second restoring force characteristic model 9B. According to the above configuration, the first restoring force characteristic model 9A represents the restoring force characteristics of the reinforced concrete building 100, and the second restoring force characteristic model 9B represents the restoring force characteristics of the reinforced concrete building 100 taking into account the slip characteristics. This makes it possible to appropriately realize parallel models in cases where the building 100 is a reinforced concrete building 100 and it is desired to take into account the slip characteristics.

[0047] In addition, the observation data acquisition unit 4 calculates the hysteretic absorbed energy Eexp at the time of observation from the relative displacement time history data Dexp at the time of observation and the load time history data Qexp at the time of observation, and the response analysis unit 6 calculates the hysteretic absorbed energy Eana at the time of analysis from the relative displacement time history data Dexp at the time of observation and the load time history data Qana at the time of analysis, and the objective function f is expressed by the residual ΔQ and the difference ΔE between the hysteretic absorbed energy Eexp at the time of observation and the hysteretic absorbed energy Eana at the time of analysis. According to the above configuration, the objective function f is expressed by the difference ΔE between the hysteretic absorbed energy Eexp at the time of observation and the hysteretic absorbed energy Eana at the time of analysis, in addition to the residual ΔQ between the load time history data Qexp at the time of observation and the load time history data Qana at the time of analysis. This makes it possible to reduce the number of solutions to be searched.

[0048] In addition, the parallel model construction unit 5 selects one value from each of the search parameters to determine multiple combinations of values ​​to be generated, generates multiple combinations, and sets the values ​​included in the combinations to construct a parallel model 8, thereby constructing multiple parallel models 8. The response analysis unit 6 performs response analysis for each of the multiple combinations using the parallel model 8 constructed corresponding to the combination, obtains the load time history data Qana at the time of analysis, and calculates the historical absorbed energy Eana at the time of analysis. The parallel model update unit 7 calculates an objective function f for each of the multiple combinations using the corresponding load time history data Qana at the time of analysis and the historical absorbed energy Eana at the time of analysis, and generates multiple new combinations of search parameter values ​​based on the combination with the smallest calculated objective function f. The parallel model construction unit 5 sets the values ​​included in the newly generated combinations to generate a parallel model 8, thereby updating the multiple parallel models 8. According to the above configuration, the parallel model construction unit 5 first generates a plurality of combinations of values ​​by selecting one value from each of the search parameters, and then sets the values ​​included in each of the plurality of combinations to construct a parallel model 8, thereby constructing a plurality of parallel models 8. Furthermore, the response analysis unit 6 performs a response analysis for each of the plurality of combinations using the parallel model 8 constructed corresponding to the combination, obtains load time history data Qana during the analysis, and calculates hysteretic absorbed energy Eana during the analysis. Then, for each of the multiple combinations, the parallel model update unit 7 calculates the objective function f using the corresponding load-time history data Qana during analysis and the hysteretic absorbed energy Eana during analysis. If there is a combination that optimizes the objective function f, it is considered that it has a value close to the combination with a small objective function f. Therefore, the parallel model update unit 7 generates multiple new combinations of search parameter values ​​based on the combination with the small calculated objective function f among the multiple combinations, and updates the multiple parallel models 8 by generating parallel models 8 by setting the values ​​included in each of the newly generated multiple combinations. Thereafter, the response analysis unit 6 performs a response analysis on each of the updated parallel models 8, obtains the load time history data Qana at the time of analysis for each of the multiple combinations, and calculates the hysteretic absorbed energy Eana at the time of analysis. In this way, by repeating the processing of the response analysis unit 6 and the parallel model update unit 7, it is possible to efficiently obtain a combination of search parameter values ​​that appropriately reduces the objective function f, and a parallel model 8 in which the combination of search parameter values ​​is set.

[0049] The building soundness assessment system 10 as described above also includes a building restoring force characteristic model estimation system 1, an earthquake response analysis unit 11 that inputs time history data of an expected earthquake (acceleration time history data Aexp) to the restoring force characteristic model estimated by the building restoring force characteristic model estimation system 1, performs response analysis, and acquires the response of the building 100 when the expected earthquake occurs, and a soundness assessment unit 12 that estimates the amount of deformation of each story of the building 100 based on the response, and assesses the soundness of the building 100 when the expected earthquake occurs based on the amount of deformation. In the above configuration, by inputting time history data corresponding to an earthquake of a scale that is likely to occur in the future as an anticipated earthquake into the restoring force characteristic model that represents the current state of the building 100 estimated by the building restoring force characteristic model estimation system 1, it is possible to determine and predict the soundness of the current state of the building 100 if this anticipated earthquake were to occur in the future. Furthermore, the restoring force characteristic model more accurately represents the current state of the building 100 by the above-described building restoring force characteristic model estimation system 1. Therefore, if a predicted earthquake occurs in the future, it is possible to more accurately determine and predict the state of the building 100 and its soundness.

[0050] (Verification example) Next, a verification carried out on the above-described building restoring force characteristic model estimation system 1 will be described. First, a scaled-down specimen simulating a 20-story reinforced concrete building 100 as shown in Fig. 1 was prepared, and a vibration experiment was conducted on the scaled-down specimen using a large shaking table. Based on the time history data observed from the experiment, a restoring force characteristic model was estimated by the building restoring force characteristic model estimation system 1 of the above embodiment. As a result, the search parameters of the estimated restoring force characteristic model, i.e., the initial stiffness (k1_t) of the first restoring force characteristic model 9A, the initial stiffness (k1_s) of the second restoring force characteristic model 9B, the yield displacement (d1_s), and the stiffness reduction rate (k2_s / k1_s), were 0.52, 0.35, 0.31, and 0.015, respectively.

[0051] FIG. 11 shows load time history data observed using a scaled-down test specimen in a vibration experiment, and load time history data obtained by inputting the same time history data as in the vibration experiment into the restoring force characteristic model estimated by the building restoring force characteristic model estimation system of the above embodiment. FIG. 12 shows the hysteretic absorbed energy observed by the scaled-down test specimen in the vibration excitation experiment, and the hysteretic absorbed energy obtained by inputting the same time history data as in the vibration excitation experiment into the restoring force characteristic model estimated by the building restoring force characteristic model estimation system of the above embodiment. FIG. 13 shows the relationship between load and deformation observed using a scaled-down test specimen in a vibration experiment, and the relationship between load and deformation obtained by inputting the same time history data as in the vibration experiment into the restoring force characteristic model estimated by the building restoring force characteristic model estimation system of the above embodiment. In either case, the results are very similar. This shows that the restoring force characteristic model estimated by the building restoring force characteristic model estimation system 1 can accurately represent the state of the building 100 realized as a scaled-down test specimen.

[0052] In the building restoring force characteristic model estimation system 1 of the above embodiment, the objective function f is calculated for a section R of several cycles, for example, about five cycles, in which the deformation (inter-story displacement) is larger than the other sections in the relative displacement time history data Dexp during observation, as described using Fig. 4. The time history data of the after-motion section, which is smaller in deformation and follows the section R targeted for the calculation of the objective function f, was input to the restoring force characteristic model by the building restoring force characteristic model estimation system 1 of the above embodiment, and the relationship between the load and the deformation was investigated. FIG. 14 shows the relationship between load and deformation observed in the after-sway section using a scaled-down test specimen in a vibration experiment, and the relationship between load and deformation obtained by inputting the same time history data as in the vibration experiment into the restoring force characteristic model estimated by the building restoring force characteristic model estimation system of the above embodiment. 14, the results for both models are very similar even in the after-swing section. This shows that the restoring force characteristic model estimated by the building restoring force characteristic model estimation system 1 can accurately represent the state of the building 100 realized as a scaled-down test specimen even in the after-swing section.

[0053] In the above embodiment, the restoring force characteristic model estimation system 1 estimates the restoring force characteristic model for each story. Here, a mass point model having 20 mass points and simulating the entire building 100 shown in Fig. 1 was constructed, and the restoring force characteristics of each story (mass point) of the mass point model were set using the restoring force characteristic model estimated by the building restoring force characteristic model estimation system 1. Then, an excitation wave was input and a simulation analysis of the earthquake response was performed. FIG. 15 is a diagram showing the relationship between the load and deformation of each story when an estimated restoring force characteristic model is set for each story of the mass system model and a simulation analysis of the earthquake response is performed. FIG. 16 is a diagram showing the maximum inter-story deformation of each story when an estimated restoring force characteristic model is set for each story of the mass system model and a simulation analysis of the earthquake response is performed. FIG. 17 is a diagram showing the maximum story shear force of each story when an estimated restoring force characteristic model is set for each story of the mass system model and a simulation analysis of the earthquake response is performed. As shown in Figures 15, 16, and 17, the estimated restoring force characteristic model matched the experimental results, demonstrating the validity of the optimized restoring force characteristics for each layer.

[0054] The building restoring force characteristic model estimation system and building health assessment system of the present invention are not limited to the above-described embodiments explained with reference to the drawings, and various other modifications are possible within the technical scope thereof. For example, in the above embodiment, the building 100 is made of reinforced concrete, but this is not limiting. In the above embodiment, the sensor provided in the building 100 is the acceleration sensor 2, and the observed time history data is acceleration time history data, but this is not limited to this. For example, the sensor provided in the building 100 may be for observing displacement, and the observed time history data may be displacement time history data. Alternatively, the sensor provided in the building 100 may be for observing velocity, and the observed time history data may be velocity time history data. In the above embodiment, the search parameters are the initial stiffness of the first restoring force characteristic model 9A, the initial stiffness of the second restoring force characteristic model 9B, the yield displacement, and the stiffness after yield, but are not limited to this. Other parameters may also be included in the search parameters. Furthermore, in the above embodiment, the parallel model 8 is constructed by arranging the first restoring force characteristic model 9A and the second restoring force characteristic model 9B in parallel, but a parallel model may be constructed by arranging other restoring force characteristic models in parallel in addition to the first restoring force characteristic model 9A and the second restoring force characteristic model 9B. In addition to this, it is possible to select and discard the configurations given in the above embodiments and modifications, or to change them to other configurations as appropriate. [Explanation of symbols]

[0055] 1 Building restoring force characteristic model estimation system 12 Soundness assessment unit 4 Observation data acquisition unit 100 Building 5 Parallel model construction section Aexp Acceleration time history data during observation (time history data) 6 Response analysis section Dexp Relative displacement time history data during observation 7 Parallel model update unit Qexp Load time history data at the time of observation 8 Parallel model Eexp Historical absorbed energy at the time of observation 9A First Restoring Force Characteristic Model Qana Load Time History Data during Analysis 9B Second restoring force characteristic model Eana Hysteresis absorption energy during analysis 10 Building Soundness Assessment System ΔQ Residual 11 Earthquake response analysis section ΔE difference

Claims

1. A building restoring force characteristic model estimation system that estimates a restoring force characteristic model of a building based on time history data observed for the building, an observation data acquisition unit that acquires relative displacement time history data at the time of observation and load time history data at the time of observation from the time history data; a parallel model construction unit that arranges a first restoring force characteristic model and a second restoring force characteristic model related to the building in parallel as springs, and constructs parallel models by setting values ​​for parameters that express the first restoring force characteristic model and the second restoring force characteristic model; a response analysis unit that inputs the relative displacement time history data at the time of observation into the parallel model, performs a response analysis, and acquires load time history data at the time of analysis; a parallel model updating unit that searches for a value of each of search parameters to be searched for when estimating the restoring force characteristic model among the parameters, such that an objective function expressed by a residual between the load time history data at the time of observation and the load time history data at the time of analysis becomes small, and updates each of the search parameters to the searched value, thereby updating the parallel model; Equipped with The process of the response analysis unit and the parallel model update unit is repeated to estimate the restoring force characteristic model. A building restoring force characteristic model estimation system.

2. the first restoring force characteristic model represents the restoring force characteristic of a reinforced concrete building, the search parameters of the first restoring force characteristic model include an initial stiffness; the second restoring force characteristic model expresses the restoring force characteristics of a reinforced concrete building in consideration of slip characteristics, The search parameters of the second restoring force characteristic model include an initial stiffness, a yield displacement, and a stiffness reduction rate after yield, which is the stiffness after yield divided by the initial stiffness of the second restoring force characteristic model.

2. The building restoring force characteristic model estimation system according to claim 1 .

3. the observation data acquisition unit calculates hysteretic absorbed energy at the time of observation from the relative displacement time history data at the time of observation and the load time history data at the time of observation; the response analysis unit calculates hysteretic absorbed energy during analysis from the relative displacement time history data during observation and the load time history data during analysis; The objective function is expressed by the residual and the difference between the historical absorbed energy at the time of observation and the historical absorbed energy at the time of analysis.

2. The building restoring force characteristic model estimation system according to claim 1 .

4. the parallel model construction unit generates a plurality of combinations of values ​​by selecting the values ​​one by one from each of the search parameters, and constructs the parallel models by setting the value included in each of the plurality of combinations; the response analysis unit performs a response analysis for each of the plurality of combinations using the parallel model constructed corresponding to the combination, acquires load time history data at the time of the analysis, and calculates hysteretic absorbed energy at the time of the analysis; The parallel model update unit calculates the objective function for each of the plurality of combinations using the corresponding load time history data at the time of the analysis and the historical absorbed energy at the time of the analysis, generates a plurality of new combinations of the values ​​of the search parameters based on a combination among the plurality of combinations for which the calculated objective function is small, and updates the plurality of parallel models by setting the value included in the combination for each of the newly generated plurality of combinations.

4. The building restoring force characteristic model estimation system according to claim 3.

5. The building restoring force characteristic model estimation system according to any one of claims 1 to 4; an earthquake response analysis unit that inputs time history data of an expected earthquake into the restoring force characteristic model estimated by the building restoring force characteristic model estimation system, performs a response analysis, and acquires a response of the building when the expected earthquake occurs; a soundness determination unit that estimates a deformation amount of each story of the building based on the response, and determines the soundness of the building when the expected earthquake occurs based on the deformation amount; A building health assessment system comprising:

Images