JP2020112445A - Earthquake information processing device - Google Patents

Abstract

To appropriately select a member of an inspection object for determining a damage level of a building with high accuracy when an earthquake occurs.SOLUTION: An acquisition unit 22 of an earthquake information processing device 100 acquires vibration data acquired from a seismometer 10 installed at each place of an object building. An initial damage probability calculation unit 26 calculates an initial damage probability of the object building on the basis of the vibration data and a model representing the object building. A probability calculation unit 30 calculates an occurrence probability of a damage level of the object building by using a Bayesian relational expression on the basis of the initial damage probability, a probability representing the damage level of each member of the object building when the damage level of the object building is given, and the probability representing the damage level of each member of the object building about respective combinations of a plurality of members of the object building. A specification unit 32 specifies members in which difference between occurrence probabilities of the damage levels is equal to or more than a threshold set in advance about the respective combinations of the plurality of members on the basis of the occurrence probabilities of the damage levels.SELECTED DRAWING: Figure 1

Description

The present invention relates to an earthquake information processing device.

BACKGROUND ART Conventionally, a method is known in which a vibration sensor is installed in a structure to record vibration data and the seismic damage risk of the structure is quantitatively determined (for example, Patent Document 1).

There is also known a soundness confirmation method for a building that enables the response of each floor of the building to be more accurately estimated based on the earthquake response information of the building obtained by a sensor installed on a limited floor. (For example, patent document 2).

In addition, a building seismic resistance evaluation system that performs response analysis of a building using an analysis model that matches the actual situation of the building in response to changes in conditions such as deterioration over time and the weight of furniture installed on each floor is known. (For example, Patent Document 3).

In addition, the microtremor of the building that is excited by wind force or traffic vibration is measured, and the vibration characteristics of the target building are identified by extracting only the vibration component of the entire building included in the measurement record. A method for evaluating the structural soundness of a portion is known (for example, Patent Document 4).

In addition, there is known a building soundness evaluation system that predicts the soundness that can occur in a building due to an earthquake (for example, Patent Document 5).

JP 9-105665 A JP 2013-195354 JP JP 2014-16249 JP JP 2003-322585 JP 2018-77104 JP

The techniques described in Patent Documents 1 to 5 above are techniques for calculating the soundness of a building after an earthquake occurs. However, the technology described in the above patent documents does not consider using the damage level information of each member of the building obtained after the occurrence of an earthquake. In Patent Document 4, the measurement is performed by an expert in a short time after the occurrence of the earthquake, but this is not realistic.

In view of the above facts, an object of the present invention is to appropriately select a member to be inspected for accurately determining a damage level of a building when an earthquake occurs.

In order to achieve the above object, the earthquake information processing apparatus of the present invention includes an acquisition unit that acquires vibration data acquired from a seismometer installed at each location of a target building, and the vibration acquired by the acquisition unit. Based on the data and the model representing the target building preset from the dynamic response analysis result for the target building, the initial damage indicating the probability of the damage level of the target building when the earthquake of the vibration data occurs. When the initial damage probability calculation unit that calculates the probability and each of the combinations of the plurality of members of the target building are given the initial damage probability calculated by the initial damage probability calculation unit and the damage level of the target building Based on the probability of representing the damage level of each member of the target building and the probability of representing the damage level of each member of the target building, using Bayes' relational expression, a combination of the damage levels of a plurality of the members Based on the probability of occurrence of the damage level calculated by the probability calculator for each of the combination of a plurality of members, the probability calculation unit that calculates the occurrence probability of the damage level of the target building, the plurality of members The earthquake information processing apparatus includes: a specifying unit that specifies, as a member to be inspected, the member in which the difference between the occurrence probabilities of the damage level for each of the combinations is equal to or more than a preset threshold value. According to the earthquake information processing apparatus of the present invention, when an earthquake occurs, it is possible to appropriately select a member to be inspected for determining the damage level of a building.

The earthquake information processing apparatus of the present invention is a setting unit that sets a prior distribution of parameters of the model representing the target building, a prior distribution of the parameters set by the setting unit, and the parameter when the parameters are given. Calculated from the product of the probability that the damage level of the target building will occur and the probability that a combination of the damage levels of the plurality of members when the damage level of the target building will occur will be calculated. A parameter calculation unit that calculates the parameters by Bayesian estimation so that the probability of the damage level found as a result of the damage investigation among the probabilities of the damage levels of the target building occurring when a combination of damage levels occurs is increased. Further, the initial damage probability calculation unit may calculate the initial damage probability when the next earthquake occurs, using the model including the parameters calculated by the parameter calculation unit. This makes it possible to properly estimate the parameter for determining the damage level of the building.

The earthquake information processing apparatus of the present invention is such that the parameters of the model representing the target building are set for each of the plurality of target buildings, and the parameters of the prior distribution of the parameters of the plurality of target building models are plural. Common in the target building of, the parameter calculation unit, for each of the plurality of target buildings, the probability of the parameter of the prior distribution, and the parameter of the parameter when the parameter of the prior distribution is given. From a product of a probability, a probability that a damage level of the target building will occur when the parameter is given, and a probability that a combination of the damage level of the member will occur when the damage level of the target building occurs. The parameter and the parameter of the prior distribution can be calculated by hierarchical Bayesian estimation so that the probability of the damage level found as a result of the damage investigation is high among the calculated damage levels. This makes it possible to properly estimate the parameters for each building for determining the damage level of the building.

The earthquake information processing apparatus of the present invention indicates the damage level of the member to be inspected of the target building and the damage level of each member of the target building when the damage level of the target building is given by the user. Based on the probability of expressing and the probability of representing the damage level of each member of the target building, using Bayes' relational expression, the occurrence probability of the damage level of the target building for the combination of the damage levels of the plurality of members. It is possible to further include a damage level calculation unit that calculates and calculates the damage level of the target building according to the probability of occurrence of the damage level of the target building. Thereby, the damage level of the building can be accurately determined by using the inspection result of the specific member after the occurrence of the earthquake.

According to the present invention, there is an effect that a member to be inspected for accurately determining a damage level of a building can be appropriately selected.

It is a block diagram which shows schematic structure of the earthquake information processing apparatus which concerns on 1st Embodiment. It is explanatory drawing for demonstrating the model of this embodiment. It is a figure which shows an example of an inspection member specific processing routine. It is a figure which shows an example of a damage level calculation processing routine. It is a block diagram which shows schematic structure of the earthquake information processing apparatus which concerns on 2nd Embodiment.

<Outline of this embodiment>

In the event of an earthquake, if people inside the building begin to evacuate all at once, the roads will be overwhelmed with people, which will prevent fire and emergency activities. For example, according to the Tokyo Metropolitan Government Handbook for Persons with Difficulty Returning Home, determine whether or not to stay at the facility within 3 hours after the disaster, and if so, keep employees as much as possible in the target building. It has been demanded. However, it is impossible for an expert such as a person in charge of a construction company to judge the damage level (or soundness) of all target buildings in such a short time. Therefore, a system is desired that allows the manager of the target building to determine the damage level of the building immediately after the earthquake.

Therefore, when an earthquake occurs, the earthquake information processing apparatus according to the present embodiment identifies the member to be inspected in the building using Bayes' relational expression based on the vibration data of the earthquake. The building manager inspects the specified inspection target member and obtains the inspection result of each member. Then, the earthquake information processing apparatus of the present embodiment acquires the inspection result of the member by the building manager, and calculates the damage level of the damaged building based on the inspection result. Thereby, the damage level of the damaged building can be calculated accurately in a short time.

Hereinafter, embodiments of the present invention will be described in detail.

<First Embodiment>

<System configuration of earthquake information processing device>

FIG. 1 is a block diagram showing an example of the configuration of the earthquake information processing apparatus 100 according to the first embodiment. As shown in FIG. 1, the earthquake information processing apparatus 100 can be functionally represented by a configuration including a plurality of seismographs 10, a reception unit 12, a computer 20, and an output unit 40.

A plurality of seismographs 10 are installed at each location of the target building. For example, each of the plurality of seismographs 10 is installed on all floors or some floors (for example, the lowest floor and the top floor) of the target building. Each of the plurality of seismographs 10 acquires acceleration of each floor when an earthquake occurs. The acceleration of each floor is an example of the vibration data of the present invention.

The reception unit 12 receives the information input by the user. The reception unit 12 is realized by, for example, a keyboard or a mouse.

The computer 20 includes a CPU (Central Processing Unit), a ROM (Read Only Memory) that stores programs for implementing each processing routine, a RAM (Random Access Memory) that temporarily stores data, and a memory as a storage unit. , A network interface and the like. As shown in FIG. 1, the computer 20 functionally has an acquisition unit 22, a model storage unit 24, an initial damage probability calculation unit 26, a probability storage unit 28, a probability calculation unit 30, and an identification unit 32. And an inspection information acquisition unit 34 and a damage level calculation unit 36.

The acquisition unit 22 acquires the acceleration acquired from the seismograph 10 installed at each location of the target building.

The model storage unit 24 stores a model f representing the target building. The model f is a function preset based on the dynamic response analysis result for the target building.

FIG. 2 shows an explanatory diagram for explaining the model f. As shown in FIG. 2, first, a structural model S of the target building A is created in advance, and a dynamic response analysis D for seismic waves Q of various seismic motions is performed. In the structural model S, strength and rigidity of each member of the target building A are considered.

The model f is generated in advance according to the dynamic response analysis result R obtained by the dynamic response analysis D. As shown in FIG. 2, when the acceleration of each floor obtained by each seismograph 10 is input to the model f, the probability that the target building is at each damage level (small, medium, large, etc.) is output. To be done. The model f is a fragility function for calculating the probability that the target building will be at or above each damage level (small, medium, large, etc.) from the vibration data.

In the present embodiment, a case where the interlayer deformation angle is calculated from the acceleration of each floor obtained by the seismograph 10 and input to the model f will be described as an example.

Based on the acceleration of each floor of the target building acquired by the acquisition unit 22 and the model f stored in the model storage unit 24, the initial damage probability calculation unit 26 determines the target when an earthquake corresponding to the acceleration occurs. An initial damage probability representing the probability of the damage level x _i of the building is calculated.

Specifically, first, the initial damage probability calculation unit 26 calculates the interlayer deformation angle of each floor from the acceleration of each floor of the target building acquired by the acquisition unit 22. Then, the initial damage probability calculating unit 26 expresses the probability P(x _i ) of the target building having each damage level (small damage, medium damage, large damage, etc.) or higher according to the following relational expression (1). To calculate.

Note that x _i in the above formula (1) represents the damage level of the target building, P(x _i ) represents the probability that the damage level is x _i or higher, and a ₁ , a ₂ ,... , D ₁ , d ₂ ,... Represents the interlayer deformation angle of each floor. Note that α and β are parameters and are calculated in advance.

It should be noted that the model f representing the predetermined function is preset from the combination of each measured value and the probability P(x _i ) of the damage level based on the dynamic response analysis result. The model f can be created using a joint probability distribution, a spline function, a neural network, or the like. The easiness of damaging a building and the easiness of damaging an actual building in a structural model, which is an analytical model of a building, do not actually match due to errors due to modeling. Therefore, the relationship between the damage level probability P(x _i ) and the model f is expressed in the form of the above equation (1).

The probability storage unit 28 stores the probability of representing the damage level of each member of the target building when the damage level x _i of the target building is given and the probability of representing the damage level of each member of the target building. .. In the present embodiment, an event in which the member j of the target building reaches the damage level k is represented by b _jk . Therefore, b ₁₁ represents the event that the member 1 is at damage level 1.

Therefore, the probability of representing the damage level k of each member j of the target building when the damage level x _i of the target building is given is, for example, P(b ₁₁ , b ₂₁ , b ₃₂ ,... |x _i ). b ₁₁ , b ₂₁ , and b ₃₂ indicate that the member 1 has the damage level 1, the member 2 has the damage level 1, and the member 3 has the damage level 2. Further, the probability of indicating the damage level k of each member j of the target building is expressed as P(b ₁₁ , b ₂₁ , b ₃₂ ,...) For example.

In addition, the probability represented by the symbol b _jk representing the event of the damage level of the member will be described below by using “b ₁₁ , b ₂₁ , b ₃₂ ”for simplifying the description. However, the subscript number differs depending on the target member and the damage level of the member. For example, when the damage level of the member 1 is 2, it is “b ₁₂ ”.

Each of these probabilities is obtained in advance by the dynamic response analysis D when setting the model f, and is stored in the probability storage unit 28.

The probability calculating unit 30 calculates the initial damage probability P(x _i ) calculated by the initial damage probability calculating unit 26 for each combination of a plurality of members of the target building and the target building's stored in the probability storage unit 28. The probability P(b ₁₁ , b ₂₁ , b ₃₂ ,... |x _i ) representing the damage level of each member of the target building when the damage level x _i is given, and the damage level of each member of the target building Based on the probability P (b ₁₁ , b ₂₁ , b ₃₂ ,...) Representing, the occurrence probability P of the damage level x _i of the target building for the combination of the damage levels of the plurality of members is calculated using the Bayes relational expression. (X _i |b ₁₁ , b ₂₁ , b ₃₂ ,...) Is calculated. The Bayesian relational expression is expressed as, for example, the following expression (2).

The occurrence probability P(x _i |b ₁₁ , b ₂₁ , b ₃₂ ,...) Of the damage level x _i of the target building obtained by the above equation (2) is the acceleration measured by a plurality of seismographs 10, It is calculated from the model f representing the target building.

However, it is possible to inspect a specific member of the damaged building and improve the estimation accuracy of the damage level of the target building in consideration of the inspection result. Specifically, if the value of b _jk representing the event of the damage level of each member is determined from the inspection result of each member that can be inspected within a predetermined time, the damage level x _{i of the} target building is calculated according to the above equation (2). It is possible to accurately estimate the occurrence probability P(x _i |b ₁₁ , b ₂₁ , b ₃₂ ,...) Of. Therefore, in the present embodiment, the specifying unit 32 specifies the member to be inspected.

Based on the occurrence probability P(x _i |b ₁₁ , b ₂₁ , b ₃₂ ,...) Of the damage level x _i calculated by the probability calculation unit 30 for each combination of a plurality of members, the specifying unit 32 For each of the combinations of a plurality of members, a member whose difference between the occurrence probabilities P (x _i |b ₁₁ , b ₂₁ , b ₃₂ ,...) Of the damage level x _i is equal to or greater than a preset threshold value, Specify as a member to be inspected. If there is no member that exceeds the threshold, inspection will not be performed.Therefore, set the threshold elastically so that the number of members to be inspected falls within the range that can be inspected. Good.

Specifically, the identifying unit 32 calculates a difference between the occurrence probabilities of the damage level x _i of the target building when the damage levels of the plurality of members are different.

For example, when the damage level of the beam member at the corner of the first floor is level 1, the damage level of this beam member is changed to level 2. In this case, when the damage level of the beam member is level 1, the probability that the damage level of the target building is "moderate" or higher is 10%, and when the damage level of the beam member is level 2, Assume that there is an 80% chance that the damage level is "moderate" or higher. In this case, since the damage level of the target building changes significantly due to the change in the damage level of the beam member at the corner of the first floor, the beam member at the corner of the first floor is specified as the member to be inspected.

Therefore, the specification unit 32, for each of combinations of a plurality of members, when changing the level of injury of members, occurrence of damage level x _i of one of the target building probability _{P (x i | b 11,} b 21, b ₃₂ ,...) and the probability P(x _i |b ₁₂ , b ₂₂ , b ₃₂ ,...) of the damage level x _i of the other target building is equal to or greater than a preset threshold value. Specify the member as the member to be inspected.

If the target building is large, it is difficult to check all the members. Therefore, a member to be inspected in the event of a disaster may be set in advance, and the member specified by the specifying unit 32 may be set as the member to be inspected. For example, if the elevator cannot be used, it is difficult to inspect the members on the higher floors, so the members on the upper floors are excluded from the inspection target members in advance. Further, conditions may be set in advance such that 10 members are to be inspected on the same floor, or 3 members are to be inspected on different floors.

Further, a member to be inspected may be preset depending on the damage level of the member, and the member specified by the specifying unit 32 may be set as the member to be inspected. If the damage of a member is obviously high, it is unlikely that the damage level of the target building will be changed even if the member is inspected again. Therefore, it is expected that the damage level of the target building can be accurately calculated by inspecting the member that is expected to be moderately damaged.

For example, a function similar to the model f representing the target building is created in advance for each member, the damage probability of the member is calculated according to the acceleration when an earthquake occurs, and the damage probability is a predetermined threshold value (for example, 0.4) or more. A certain member is preset as a member to be inspected. Then, among the preset members, the member specified by the specifying unit 32 is set as the inspection target member.

The output unit 40 outputs each of the members to be inspected, which are specified by the specifying unit 32. The output unit 40 is realized by, for example, a display or the like. The inspector of the target building, which is the user, confirms the member to be inspected displayed on the output unit 40.

The user goes to the location of each member to be inspected displayed on the output section 40 to check the damage level of each member. Then, the user inputs the inspection result indicating the damage level of each member to be inspected to the computer 20 via the reception unit 12.

The inspection information acquisition unit 34 acquires the inspection result of each member received by the reception unit 12 from the inspector.

The damage level calculation unit 36 calculates the damage level of each member to be inspected of the target building input by the user, the probability of representing the damage level of each member of the target building when the damage level of the target building is given, and the target Based on the probability representing the damage level of each member of the building and Bayes' relational expression, the probability of occurrence of the damage level of the target building for the combination of the damage levels of multiple members is calculated, and the damage level of the target building The damage level of the target building is calculated according to the occurrence probability of.

Specifically, first, the damage level calculation unit 36 calculates the damage level x _{i of the} target building based on the damage level of each member representing the inspection result acquired by the inspection information acquisition unit 34 according to the above equation (2). The occurrence probability P(x _i |b ₁₁ , b ₂₁ , b ₃₂ ,...) Is calculated again. In addition, when the occurrence probability of the damage level x _i of the target building is calculated according to the above equation (2), the events b ₁₁ , b ₂₁ , b ₃₂ representing the initial damage level of each member are, for example, inspection results. Accordingly, b ₁₂ , b ₂₂ , and b ₃₂ are obtained.

Next, the damage level calculation unit 36 is based on the occurrence probability P(x _i |b ₁₂ , b ₂₂ , b ₃₂ ,...) Of the damage level x _i of the target building calculated again by the damage level calculation unit 36. Then, the probability Q(x _i ) that the damage level of the target building is the damage level x _i is calculated according to the following formula (3). The probability of occurrence of the damage level x _i of the target building represents the probability of being above the damage level (for example, the probability of being above "medium damage"). Therefore, the probability that the damage level of the target building is the damage level x _i (for example, the probability that the damage level is “moderate”) is calculated according to the following formula (3). For example, according to the following formula (3), the probability of "medium damage" is obtained by subtracting the probability of "major damage" or more from the probability of "medium damage" or more. _I having the maximum probability Q(x _i ) calculated by the following equation (3) is the final damage level determination result of the target building.

It should be noted that n in the above formula (3) represents the total number of damage levels, and when the damage levels are “small damage”, “medium damage”, and “major damage”, n=3.

When the damage level calculation unit 36 calculates the occurrence probability P(x _i |b ₁₁ , b ₂₁ , b ₃₂ ,...) Of the damage level x _i of the target building, the formula (2) is used instead. Then, the following equation (4) may be used. When comparing the occurrence probabilities of the respective damage levels x _i of the target building in a state where the damage level of the inspected member is fixed, the denominator on the right side of the above equation (2) is common and therefore need not be calculated. Therefore, the occurrence probability of the damage level x _i of the target building can also be calculated by the following formula (4).

The output unit 40 outputs the final damage level of the target building calculated by the damage level calculation unit 36 as a result. The administrator of the target building, which is the user, confirms the result displayed on the output unit 40 and grasps the damage level of the target building.

<Operation of earthquake information processing device>

Next, the operation of the earthquake information processing apparatus 100 will be described with reference to FIGS. 3 and 4. The earthquake information processing apparatus 100 executes the inspection member identification processing routine of FIG. 3 and the damage level calculation processing routine of FIG.

<Inspection member identification processing routine>

When an earthquake occurs and the accelerations are measured by the plurality of seismographs 10, the earthquake information processing apparatus 100 executes the inspection member identification processing routine shown in FIG.

In step S100, the acquisition unit 22 acquires the acceleration acquired from the seismograph 10 installed at each location of the target building.

In step S102, the initial damage probability calculation unit 26 reads the model f stored in the model storage unit 24. Then, in step S102, the initial damage probability calculator 26 calculates the initial damage probability P(x) according to the above equation (1) based on the acceleration of each floor of the target building acquired in step S100 and the read model f. _i ) is calculated.

In step S104, the probability calculation unit 30 indicates the probability P(b ₁₁ , b ₂₁ , b ₂₁₎ representing the damage level of each member of the target building when the damage level x _i of the target building stored in the probability storage unit 28 is given. b ₃₂ ,... |x _i ) and the probability P(b ₁₁ , b ₂₁ , b ₃₂ ,...) Representing the damage level of each member of the target building are read. Then, in step S104, the probability calculating unit 30 calculates the initial damage probability P(x _i ) calculated in step S102 and the read probability P(b ₁₁ , b) for each of the combinations of the plurality of members of the target building. ₂₁ , b ₃₂ ,... |x _i ) and the read probability P(b ₁₁ , b ₂₁ , b ₃₂ ,...) Based on the above equation (2), the damage level x of the target building x _i probability of occurrence _{P (x i | b 11,} b 21, b 32, ···) is calculated.

In step S106, the identifying unit 32 combines a plurality of members based on the occurrence probability P(x _i |b ₁₁ , b ₂₁ , b ₃₂ ,...) Of the damage level x _i calculated in step S104. For each of the above, the occurrence probability P(x _i |b ₁₁ , b ₂₁ , b ₃₂ ,...) Of the damage level x _i of one target building when the damage level of the member is changed and the other target building The member whose difference from the occurrence probability P(x _i |b ₁₂ , b ₂₂ , b ₃₂ ,...) Of the damage level x _i of No. 1 is equal to or greater than a preset threshold is specified as a member to be inspected.

In step S108, the identification unit 32 outputs the member to be inspected identified in step S106 as a result.

The output unit 40 outputs each of the members to be inspected, which are specified by the specifying unit 32.

The user confirms the member to be inspected displayed on the output unit 40. Next, the user goes to the location of each member to be inspected displayed on the output unit 40 to check the damage level of each member. Then, the user inputs the inspection result indicating the damage level of each member to be inspected to the computer 20 via the reception unit 12. When the inspection result is input to the computer 20, the earthquake information processing apparatus 100 executes the damage level calculation processing routine shown in FIG.

<Damage level calculation processing routine>

In step S200, the inspection information acquisition unit 34 acquires the inspection result of each member from the inspector accepted by the acceptance unit 12.

In step S202, the damage level calculation unit 36 calculates the occurrence level P() of the damage level x _i of the target building based on the damage level of each member representing the inspection result acquired in step S200 according to the above equation (2). x _i |b ₁₁ , b ₂₁ , b ₃₂ ,...) Is calculated again. Then, in step S202, the damage level calculation unit 36, based on the recalculated probability P (x _i |b ₁₂ , b ₂₂ , b ₃₂ ,...) Of the damage level x _i of the target building, According to the equation (3), the probability Q(x _i ) that the damage level of the target building is the damage level x _i is calculated.

In step S204, the damage level calculation unit 36 determines _i having the maximum probability Q(x _i ) calculated in step S202 as the final damage level of the target building.

In step S206, the damage level calculation unit 36 outputs the damage level obtained in step S204 as a result.

The output unit 40 outputs the final damage level of the target building calculated by the damage level calculation unit 36 as a result. The administrator of the target building, which is the user, confirms the result displayed on the output unit 40 and grasps the damage level of the target building.

As described in detail above, the earthquake information processing apparatus according to the present embodiment is a target set in advance based on the acceleration acquired from the seismograph installed at each location of the target building and the dynamic response analysis result for the target building. An initial damage probability representing the probability of the damage level of the target building is calculated based on the model representing the building. Then, the earthquake information processing apparatus, for each combination of a plurality of members of the target building, the initial damage probability, the probability of representing the damage level of each member of the target building when the damage level of the target building is given, Based on the probability representing the damage level of each member of the building, Bayes' relational expression is used to calculate the occurrence probability of the damage level of the target building for the combination of the damage levels of the plurality of members. Then, the earthquake information processing apparatus presets the difference between the damage level occurrence probabilities for each of the plurality of member combinations based on the damage level occurrence probabilities calculated for each of the plurality of member combinations. The members that exceed the threshold value are specified as the members to be inspected. Accordingly, when an earthquake occurs, it is possible to appropriately select the member to be inspected for determining the damage level of the building. Moreover, the damage level of the building can be accurately determined by using the inspection result of the specific member after the earthquake.

In addition, it is possible to specify a member in the target building to be inspected by using the Bayesian relational expression and present a highly accurate determination result at an early stage.

In addition, when an earthquake occurs, it is possible to immediately determine the damage level of the target building and appropriately determine whether to evacuate from the target building or stay within the target building.

<Second Embodiment>

Next, a second embodiment will be described. In the second embodiment, when the damage level of the target building is determined by a diagnosis of an expert or the like after the occurrence of the earthquake, the first embodiment is that the damage level is used to update the parameters α and β of the above formula (1). Different from the form. Regarding the earthquake information processing apparatus according to the second embodiment, the same components as those of the earthquake information processing apparatus 100 according to the first embodiment are designated by the same reference numerals, and detailed description thereof will be omitted.

<System configuration of earthquake information processing device>

FIG. 5: is a block diagram which shows an example of a structure of the earthquake information processing apparatus 200 which concerns on 2nd Embodiment. As shown in FIG. 5, the earthquake information processing apparatus 200 can be functionally represented by a configuration including a plurality of seismographs 10, a reception unit 12, a computer 220, and an output unit 40.

The computer 220 includes a CPU (Central Processing Unit), a ROM (Read Only Memory) that stores programs for implementing each processing routine, a RAM (Random Access Memory) that temporarily stores data, and a memory as a storage unit. , A network interface and the like. As shown in FIG. 5, the computer 220 functionally has an acquisition unit 22, a model storage unit 24, an initial damage probability calculation unit 26, a probability storage unit 28, a probability calculation unit 30, and an identification unit 32. An inspection information acquisition unit 34, a damage level calculation unit 36, a setting unit 38, and a parameter calculation unit 39 are provided.

Assuming that the prior probability parameters α and β of the above equation (1) follow a certain distribution, using the super prior distributions P(α) and P(β), the above equation (4) becomes It can be described as 5).

After the earthquake, when the phenomenon b _jk representing the damage level of each member of the target building is clarified by the expert, the parameters α and β that maximize the above equation (5) are calculated using the hierarchical Bayesian estimation. By obtaining it, the accuracy of the initial damage probability of the above formula (1) can be improved.

Therefore, in the second embodiment, the parameters α and β of the model f representing the target building are estimated based on the damage level of each member of the target building confirmed in the previous earthquake, and when the next and subsequent earthquakes occur. It is used to calculate the initial damage probability of.

The setting unit 38 sets the prior distribution of the parameters α and β of the model f representing the target building. For example, the setting unit 38 models the prior distribution with a normal distribution. When the prior distribution is modeled by a normal distribution, for example, P(α)=F(0,1) and P(β)=F(1,1) can be set.

After the earthquake, experts will investigate the damage to the target building. By this damage investigation, the damage level (for example, x ₁ or x ₂ ) of the target building and the damage level (for example, b ₁₁ or b ₁₂ ) of each member are determined. The first term on the right side of the above equation (5) is determined based on the determined damage level of the target building and the damage level of each member. The damage level of the target building and the damage level of each member determined by the damage investigation are stored in a predetermined storage unit (not shown) of the computer 20.

The parameter calculation unit 39 calculates the parameter prior distributions P(α) and P(β) set by the setting unit 38, and the probability P(x _i ) of the damage level x _i of the target building when the parameters are given. |α, β) and the probability P(b ₁₁ , b ₂₁ ,... |x _i ) of a combination of damage levels of a plurality of members when the damage level x _i of the target building occurs Equation (3) is used from the calculated probability P(x _i |b ₁₁ , b ₂₁ ,...) Of the damage level x _i of the target building when a combination of damage levels of a plurality of members occurs. Then, the probability of each damage level X _i is obtained, and the parameters α and β are calculated by Bayesian estimation so that the probability of the damage level found as a result of the damage investigation becomes the maximum.

Specifically, the parameter calculation unit 39 substitutes the left side of the above equation (5) into the equation (3) using MAP estimation for the parameters α and β, and calculates Q(x _i ) as follows: The above equation (1) is updated by using the parameters α and β when the probability of the damage level found as a result of the damage investigation becomes the highest as new parameters.

The initial damage probability calculation unit 26 of the second embodiment calculates the initial damage probability P(x _i ) when the next earthquake occurs, using the model f including the parameters α and β calculated by the parameter calculation unit 39. To do.

As described in detail above, the earthquake information processing apparatus of the second embodiment sets the prior distribution of the parameters of the model representing the target building, and the prior distributions of the parameters P(α) and P(β) probability _{P (x i | α, β} ) to the level of injury x _i of the target building when given occurs with the combination of the level of injury of a plurality of members occurs when the damage level x _i of the target building occurs Probability P (the probability P(b ₁₁ ,b ₂₁ ,... |x _i ) of occurrence of damage level x _i of the target building when a combination of damage levels of a plurality of members occurs, calculated from the product of probability P(b ₁₁ , b ₂₁ ,... |x _i ). The parameters α and β are calculated by Bayesian estimation so that the probability of the damage level found as a result of the damage survey is large among the damage probabilities Q(x _i ) calculated from x _i |b ₁₁ , b ₂₁ ,. To do. Then, the earthquake information processing apparatus of the second embodiment uses the model f including the calculated parameters α and β to calculate the initial damage probability P(x _i ) when the next earthquake occurs. This makes it possible to properly estimate the parameter for determining the damage level of the building. Furthermore, the accuracy of the next damage level determination can be improved based on the damage inspection result of the target building.

<Third Embodiment>

Next, a third embodiment will be described. The third embodiment differs from the second embodiment in that the parameters α and β of the above equation (1) are set for each building. Since the configuration of the third embodiment is the same as that of the second embodiment, the same reference numerals are given and the description thereof is omitted.

Note that the earthquake information processing apparatus of the third embodiment is in charge of the organization that owns or manages a plurality of target buildings that improve the accuracy of the data when similar damage level data of the target buildings is collected. Used by the person.

In the third embodiment, parameters α _m and β _{m of the} model f representing the target building are set for each of the plurality of target buildings m. Further, the parameters σ ₁ and σ ₂ of the prior distribution of the parameters α _m and β _m of the model f of the target buildings m are set to be common to the target buildings m.

In the third embodiment, the parameters α _m and β _m of the above equation (1) are set for each target building, and it is considered that the parameters follow a certain distribution common to all target buildings.

The above equation (1) represents the difference between the probability of representing the damage level obtained from the numerical analysis and the probability of representing the actual damage level, but as in the first embodiment or the second embodiment, all the parameters are There is a case where the same thing is common to all target buildings as a constraint condition. However, it is also unreasonable to set the parameters of the above formula (1) completely for each target building, and it is considered that there is some degree of commonality among the target buildings.

Therefore, in the third embodiment, the above equation (1) is transformed into the following equation (6). It can be assumed that m in the following equation (6) represents the target building, and the parameters α _m and β _m follow a normal distribution common to the target building.

The above equation (5) can be rewritten as shown in the following equation (6) using the hyper prior distributions of the parameters α _m and β _m .

Based on each data, the parameters α _m and β _m that maximize the posterior probability of Q(x _i ) and the parameters σ ₁ and σ ₂ of the prior distribution are calculated according to the above formulas (7) and (3). The accuracy of calculation of the initial damage probability of the target building can be improved. Note that x _i , b ₁₁ , b ₂₁ , b ₃₂ ... In the above formula (7) are data determined based on the results of a survey by an expert.

Therefore, the parameter calculation unit 39 of the third embodiment calculates the probabilities P(σ ₁ ) and P(σ ₁ ) of the parameters σ ₁ and σ ₂ of the prior distribution for each of the plurality of target buildings m according to the above equation (7). σ ₂ ), the probability P(α _m |σ ₁ ), P(β _m |σ ₂ ) of the parameters when the parameters σ ₁ and σ _{2 of} the prior distribution are given, and when the parameters are given, probability _{P (x i | α, β} ) of the building damage level _{x i} occurs and the probability combinations damage level of member occurs when the damage level _{x i} of the target building occurs _{P (b 11,} b ₂₁ , ||x _i ), the parameters α _m and β _m and the parameters σ ₁ and σ _{2 of the} prior distribution are calculated by hierarchical Bayesian estimation.

As described in detail above, in the earthquake information processing apparatus of the third embodiment, the parameters α _m and β _{m of the} model f representing the target building are set for each of the plurality of target buildings m. The parameters σ ₁ and σ ₂ of the prior distribution of the parameters α and β of the target building model f are common to a plurality of target buildings. The earthquake information processing apparatus according to the third embodiment includes the probability P(σ ₁ ), P(σ ₂ ) of the parameters σ ₁ , σ ₂ of the prior distribution and the parameters of the prior distribution for each of the plurality of target buildings m. The probability P(α _m |σ ₁ ), P(β _m |σ ₂ ) of the parameters when given, and the probability P(x _i |) of the damage level x _i of the target building when the parameters are given. Q calculated from the product of α, β) and the probability P(b ₁₁ , b ₂₁ ,... |x _i ) of the combination of the damage levels of the members when the damage level x _i of the target building occurs The parameters α _m and β _m and the parameters σ ₁ and σ _{2 of the} prior distribution are calculated by hierarchical Bayesian estimation so that the probability of the damage level found in the damage investigation in (x _i ) is increased. This makes it possible to properly estimate the parameters for each building for determining the damage level of the building. Further, based on the data of the inspection results regarding the damage of a plurality of target buildings, it is possible to improve the accuracy of determination of the damage level when the next earthquake occurs.

The present invention is not limited to the above-described embodiments, and various modifications and applications can be made without departing from the spirit of the present invention.

For example, in the above-described embodiment, P(x _i ) is set as the damage level of the entire target building, but this is set as the damage level of each floor or the damage level of each member, and the determination criterion for the emergency risk degree determination is separately provided. The damage level (or soundness) of the entire target building may be determined based on the above.

Further, in the above-described embodiment, the case of using the above formula (1) is described as an example, but the present invention is not limited to this. For example, the following equation may be used as an equation representing the relationship between the probability P(x _i ) and the model f.

Further, although the above has described the mode in which the program according to the present invention is stored (installed) in the storage unit (not shown) in advance, the program according to the present invention may be a CD-ROM, a DVD-ROM, a micro SD card, or the like. It is also possible to provide it in the form recorded on the recording medium.

10 seismograph 12 reception unit 20, 220 computer 22 acquisition unit 24 model storage unit 26 initial damage probability calculation unit 28 probability storage unit 30 probability calculation unit 32 identification unit 34 inspection information acquisition unit 36 damage level calculation unit 38 setting unit 39 parameter calculation Unit 40 Output unit 100, 200 Earthquake information processing device

Claims (4)

An acquisition unit that acquires vibration data acquired from seismographs installed at each location of the target building,
Based on the vibration data acquired by the acquisition unit and a model representing the target building preset from a dynamic response analysis result for the target building, the target building when the earthquake of the vibration data occurs. An initial damage probability calculation unit that calculates an initial damage probability that represents the probability of the damage level of
For each combination of a plurality of members of the target building, the initial damage probability calculated by the initial damage probability calculation unit, and the damage level of each member of the target building when the damage level of the target building is given. Based on the probability of expressing and the probability of representing the damage level of each member of the target building, using Bayes' relational expression, the occurrence probability of the damage level of the target building for the combination of the damage levels of the plurality of members. A probability calculating unit for calculating,
Based on the damage level occurrence probability calculated by the probability calculating unit for each of the plurality of member combinations, a difference between the damage level occurrence probabilities for each of the plurality of member combinations is set in advance. A member that is equal to or greater than the specified threshold value, and a specifying unit that specifies the member as an inspection target,
Earthquake information processing equipment including. A setting unit that sets the prior distribution of the parameters of the model representing the target building,
Prior distribution of the parameter set by the setting unit, the probability that the damage level of the target building will occur when the parameter is given, and the plurality of the members when the damage level of the target building occurs Calculated from the product of the probability of occurrence of the combination of damage levels, the probability of occurrence of damage level of the target building when a combination of the damage levels of the plurality of members occurs, found as a result of damage investigation To further increase the probability of damage level, further comprising a parameter calculation unit that calculates the parameter by Bayesian estimation,
The initial damage probability calculation unit uses the model including the parameters calculated by the parameter calculation unit to calculate the initial damage probability when the next earthquake occurs,
The earthquake information processing apparatus according to claim 1. For each of the plurality of target buildings, the parameters of the model representing the target building are set, the parameters of the prior distribution of the parameters of the model of the plurality of target buildings are common in the plurality of target buildings,
The parameter calculation unit is provided with the probability of the parameter of the prior distribution, the probability of the parameter when the parameter of the prior distribution is given, and the parameter for each of the plurality of target buildings. Damage among the damage levels calculated from the product of the probability that the damage level of the target building sometimes occurs and the probability that a combination of the damage levels of the members when the damage level of the target building occurs occurs Calculate the parameter and the parameter of the prior distribution by hierarchical Bayesian estimation so that the probability of damage level found as a result of the investigation is increased,
The earthquake information processing apparatus according to claim 2. The damage level of the member to be inspected of the target building input by the user, the probability of representing the damage level of each member of the target building when the damage level of the target building is given, and each of the target buildings Based on the probability representing the damage level of the member, by using the Bayesian relational expression, calculate the probability of occurrence of the damage level of the target building for the combination of the damage levels of the plurality of members, the damage level of the target building Further comprising a damage level calculation unit that calculates a damage level of the target building according to the occurrence probability of
The earthquake information processing apparatus according to any one of claims 1 to 3.