JP6930901B2 - Vibration control control system, vibration control control method, vibration analysis device and vibration analysis method - Google Patents

Description

The present invention relates to a vibration damping control system, a vibration damping control method, a vibration analysis device, and a vibration analysis method.

Vibration damping technology in a vibration damping control system includes a passive method and an active method. In the active method, the actuator that applies stress to the object is controlled based on the measured value by the sensor provided on the object to actively suppress the vibration of the object. As an active vibration control control system in the field of construction, a control theory that treats a controlled object as a stationary and linear system is known.

Japanese Unexamined Patent Publication No. 2005-249687

However, in the case of an actual building structure, disturbances such as seismic motion that causes stress in the building structure are unsteady, and the building structure itself has non-linear characteristics. Therefore, even if the control theory that treats the controlled object as a steady and linear system is applied to the vibration damping control system of the building structure, the vibration damping effect as expected may not be obtained.

The problem to be solved by the present invention is to provide a vibration damping control system, a vibration damping control method, a vibration analysis device, and a vibration analysis method capable of suppressing vibration of a building structure by a simpler method.

The vibration damping control system of one aspect of the present invention includes a first artificial neural network (hereinafter referred to as a first NN) that has been trained using a vibration simulator that simulates the vibration of a building structure having a plurality of layers. This is a vibration damping control system including a control unit that adjusts a force that a stress adjusting unit acts as a horizontal even force on two floor portions sandwiching a specific layer among the plurality of layers by the first NN.

The vibration suppression control system includes a detection unit that detects the movement of the building structure, and a stress adjustment unit that applies an even force to the two floor portions in the building structure based on the control. The first NN of the control unit controls the stress adjustment unit based on a physical quantity representing the movement of the building structure.

In the vibration suppression control system, the second artificial neural net of the vibration analysis device that analyzes the vibration based on the virtual building structure model by the response analysis that drives the virtual building structure model of the building structure is stressed. A control amount of the virtual stress adjustment unit that replaces the adjustment unit is generated, and based on the generated control amount, between the first quality point and the second quality point of the virtual building structure model corresponding to the two floor parts. In, or between the first mass point and the reference plane of the virtual building structure model, the virtual stress adjusting unit applies a horizontal even force to drive the virtual building structure model. This is carried out to generate state values representing the movements of each part of the virtual building structure model, and the feature information of the first NN of the control unit is the degree of vibration control based on the state values generated by the vibration simulator. It is determined by learning with the reward.

The vibration analysis device of one aspect of the present invention is a vibration analysis device that simulates the vibration of a building structure including a plurality of layers, and is a plurality of vibration analysis devices based on a control amount generated by an artificial neural net (first NN). The virtual building structure model of the building structure is driven so that the stress adjusting portion acts a horizontal even force on the two floor portions sandwiching the specific layer of the layers, and the virtual building structure model of the building structure is operated. It is a vibration analysis device including a vibration simulator that generates a state value representing the movement of each part by vibration analysis.

The vibration simulator in the vibration analysis device of one aspect of the present invention dynamically adjusts the couple acting on the two floor portions in the building structure to generate the state value.

In the vibration damping control method of one aspect of the present invention, an artificial neural net (first NN) that has been trained using a simulator that simulates vibration of a building structure having a plurality of layers is one of the plurality of layers. This is a vibration damping control method including a process of adjusting a couple in the horizontal direction caused by a stress adjusting unit acting on two floor portions sandwiching a specific layer.

The vibration analysis method of one aspect of the present invention is a vibration analysis method of a vibration analysis device that simulates the vibration of a building structure having a plurality of layers, and is based on a control amount generated by an artificial neural net (first NN). , The virtual building by driving the virtual building structure model of the building structure so that the stress adjusting unit acts a couple in the horizontal direction on the two floor portions sandwiching the specific layer among the plurality of layers. This is a vibration analysis method that includes a process of performing response analysis that generates state values that represent the movement of each part of the structure model.

According to the present invention, it is possible to provide a vibration damping control system, a vibration damping control method, a vibration analysis device, and a vibration analysis method capable of suppressing vibration of a building structure by a simpler method.

It is a figure for demonstrating the outline of the vibration damping control system which concerns on embodiment of 1st Embodiment. It is a figure which shows a part of the building where the vibration damping device of an embodiment is installed. It is sectional drawing which shows the vibration damping device of embodiment. It is a block diagram of the control system of the vibration damping control system 1 of embodiment. It is a block diagram of the machine learning device 10 of an embodiment. It is a figure for demonstrating the virtual building structure model of embodiment. It is a schematic diagram which shows the model of the neuron of an embodiment. It is a schematic diagram which shows the neural network of an embodiment. It is a flowchart which shows the procedure of the reinforcement learning process of an embodiment.

Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings. In the following description, the same reference numerals are given to configurations having the same or similar functions. Then, the duplicate description of those configurations may be omitted.
In the embodiment, the building is an example of a building structure, and may have a rigid frame structure, for example. The "layer" of a building is a range sandwiched between a plurality of floors (floor portions) having different ground clearances. For example, when the plurality of floors are arranged adjacent to each other, each floor portion corresponds to the floor of the building. It should be noted that another floor on which the couple does not act may be arranged between the pair of floors on which the couple acts. Further, the "floor portion" on which a couple acts corresponds to a mass point or a reference plane when the virtual building structure model in the vibration simulator is formed by a multi-mass point system. The above reference plane corresponds to the ground surface, foundation, first floor, etc. of the building corresponding to the virtual building structure model. In the embodiment, the "feature information of the artificial neural network" is the information acquired as the vibration control rule in the artificial neural network.

(First Embodiment)
FIG. 1 is a diagram for explaining an outline of a vibration damping control system according to an embodiment of the present invention.

[Overview of damping control system]
The vibration damping control system 1 controls the vibration of the building BL with the building BL as the control target.

As shown in FIG. 1, the vibration damping control system 1 is provided, for example, in a high-rise or super-high-rise building (building structure) BL, and when an earthquake motion occurs (hereinafter, simply referred to as “vibration generation”). Suppress the shaking of the building BL. The vibration damping control system 1 is, for example, an active vibration damping control system provided as a countermeasure against long-period ground motion in a building BL. The vibration damping control system 1 of the present embodiment includes a plurality of vibration damping devices 11 (stress adjusting unit), a plurality of sensors 12 (detection unit), and a control unit 13. The vibration damping device 11 and the sensor 12 may be only one each. The building BL may have one or more layers. For example, the vibration damping control system 1 may be applied not only to a high-rise building BL but also to a low-rise (for example, 3 to 5 floors) building BL. For example, in the building BL, the structure of the layer in which the vibration damping device 11 is arranged is the same as the structure of the layer in which the vibration damping device 11 is not arranged within a predetermined range. The building BL shown in FIG. 1 has a rigid frame structure and is constructed on the ground. The same within the above-mentioned predetermined range means that the structure of the layer in which the vibration damping device 11 is arranged and the structure of the layer in which the vibration damping device 11 is not arranged have the same structural mode.

The vibration damping device 11 is arranged, for example, in the lower part (lower floor) of the building BL. The vibration damping device 11 includes a rotary damper 23 described later, and exerts a damping force when vibration is generated. The plurality of vibration damping devices 11 are arranged separately in, for example, a plurality of layers (floors) of the building BL. The location where the vibration damping device 11 is installed is not limited to the lower layer of the building BL, and may be any layer (floor).

The sensor (vibration sensor) 12 is arranged in an arbitrary layer of the building BL and detects information on the vibration of the building BL. The sensor 12 is, for example, an acceleration sensor that detects the vibration of the building BL as an acceleration, but is not limited to this. For example, the sensor 12 detects the vibration at the installed point and outputs it by dividing it into components in the three orthogonal axes.

The sensor 12 is installed, for example, in a layer on the lower layer side (including the foundation of the building and the basement) than the layer in which the vibration damping device 11 is arranged. By arranging the sensor 12 in a layer on the lower layer side than the layer on which the vibration damping device 11 is arranged, it becomes easier to perform control that reflects the state of shaking on the lower layer side. The position where the sensor 12 is arranged in each layer is not limited and may be appropriately selected. In the following description, for example, a case where the sensor 12 is arranged on the floor of the layer, that is, the floor will be described as an example.

Instead of the sensor 12 installed on the lower layer side than the layer on which the vibration damping device 11 is arranged, a sensor (not shown) may be installed on the site of the building BL (hereinafter, referred to as the ground of the building BL). It may be installed on both the floor on the lower floor side than the layer on which the vibration damping device 11 is arranged and the ground of the building BL. For example, the plurality of sensors 12 are divided and arranged on a plurality of floors of the building BL. The plurality of sensors 12 are installed in the upper part (higher floor) of the building BL, the middle layer part (middle floor), the lower part (lower floor than the layer in which the vibration damping device 11 is arranged), or the ground of the building BL, respectively. You may. The position of the layer on which the plurality of sensors 12 are arranged may be arranged at a position where the vibration characteristics of the building BL can be easily detected due to the structure of the building BL.

The control unit 13 is communicably connected to the plurality of vibration damping devices 11 and the plurality of sensors 12 via wire or wireless. The control unit 13 acquires the detection result of the sensor 12 from the sensor 12. Then, the control unit 13 calculates the magnitude and timing of the force generated in the vibration damping device 11 based on the detection result of the sensor 12. Then, the control unit 13 controls the vibration damping device 11 based on the information indicating the magnitude and timing of the force obtained by the calculation. For example, the control unit 13 adjusts the force acting on the two floor portions sandwiching the specific layer. The method is to adjust the force acting on the two floor portions sandwiching the above-mentioned specific layer based on the vibration data showing the vibration of the ground of the building BL or the vibration data showing the vibration of the building BL. May be good. For example, the vibration data of the vibration of the building BL is based on the vibration data of the vibration including the vibration data of the vibration on the lower layer side (the layer on the lower layer side) than the layer on which the vibration damping device 11 is arranged, which will be described later. It may be the one that adjusts the force acting on the two floor portions that sandwich it.

The control unit 13 is, for example, a software function unit realized by executing a program by the processor of the vibration damping control system 1. However, the control unit 13 may be realized by hardware such as LSI (Large Scale Integration), ASIC (Application Specific Integrated Circuit), FPGA (Field-Programmable Gate Array), or the software function unit and the hardware may be combined. It may be realized by collaborating.

Next, the vibration damping device 11 of the present embodiment will be described.
FIG. 2 is a diagram showing a part of the building BL in which the vibration damping device 11 of the embodiment is installed. As shown in FIG. 2, the building BL is, for example, the first beam B1, the second beam B2, the first pillar P1, the second pillar P2, and Y as a part of the structural material forming one floor of the building BL. It has a shape brace Br.

The first beam B1 extends in a substantially horizontal direction and forms a part of the floor portion F of the floor on which the vibration damping device 11 is installed. The second beam B2 is arranged substantially parallel to the first beam B1 and extends in a substantially horizontal direction, and forms a part of the ceiling portion C of the floor on which the vibration damping device 11 is installed. The first pillar P1 and the second pillar P2 extend in substantially vertical directions, respectively, and extend over the first beam B1 and the second beam B2. The vibration damping device 11 is arranged, for example, in the space S surrounded by the first beam B1, the second beam B2, the first pillar P1, and the second pillar P2.

The Y-shaped brace Br has a first brace Br1, a second brace Br2, and a connecting portion Br3. The first end portion Br1a of the first brace Br1 is connected to the joint portion between the second beam B2 and the first pillar P1. The second end portion Br1b of the first brace Br1 is located obliquely downward with respect to the first end portion Br1a, and is located at a substantially central portion in the horizontal direction of the space S. Similarly, the first end portion Br2a of the second brace Br2 is connected to the joint portion between the second beam B2 and the second pillar P2. The second end portion Br2b of the second brace Br2 is located obliquely downward with respect to the first end portion Br2a, and is located at a substantially central portion in the horizontal direction of the space S. The connecting portion (top) Br3 connects the second end portion Br1b of the first brace Br1 and the second end portion Br2b of the second brace Br2.

As shown in FIG. 2, in the present embodiment, the vibration damping device 11 is installed between the connecting portion Br3 of the Y-shaped brace Br and the second pillar P2. The vibration damping device 11 is fixed to the connecting portion Br3 of the Y-shaped brace Br and the second pillar P2, respectively. The term "fixed to XX" as used in the present application is not limited to the case of being directly fixed to "XX", but also includes the case of being indirectly fixed by interposing another member. In other words, "fixed to XX" means "fixed relative to XX". The Y-shaped brace Br is an example of the "first member". The second pillar P2 is an example of the “second member”. The Y-shaped brace Br and the second pillar P2 are examples of a set of members that are relatively displaced (relatively deformed) when vibration is generated. The structure of the building BL in which the vibration damping device 11 is provided is not limited to the above example.

Next, the configuration of the vibration damping device 11 of the present embodiment will be described.
FIG. 3 is a cross-sectional view showing the vibration damping device 11 of the embodiment. As shown in FIG. 3, the vibration damping device 11 includes a case 21, a fixing member 22, a rotary damper 23, a fixing mechanism 24, a transmission unit 25, and an actuator 26.

The case 21 is installed on the floor portion F formed by the first beam B1. The case 21 is formed in a box shape and houses a part of the rotary damper 23, a second fixing portion 52 of the fixing mechanism 24, a transmission portion 25, and an actuator 26. The case 21 has an opening 21a through which the rotary damper 23 is passed. The case 21 is not an essential component and may be omitted.

The fixing member (mounting member, connecting member) 22 is provided between the case 21 and the second pillar P2. The fixing member 22 is fixed to the second pillar P2 (for example, the leg portion of the second pillar P2), and is integrally displaced with the second pillar P2 (for example, the leg portion of the second pillar P2) when vibration is generated. The case 21, the second fixing portion 52 of the fixing mechanism 24, and the actuator 26 are fixed to the fixing member 22.

The rotary damper 23 is arranged between the connecting portion Br3 of the Y-shaped brace Br and the fixing member 22. Here, the "axial direction AD" and the "diameter direction R" of the rotary damper 23 are defined. The axial direction AD is the axial direction (longitudinal direction) of the shaft member 36 described later of the rotary damper 23. The radial direction R is a direction substantially orthogonal to the axial direction AD, for example, a direction radially separated from the shaft member 36. Further, "rotation" in the following description means rotation around a center line extending in the axial direction AD.

In the present embodiment, the rotary damper 23 is arranged so that the axial direction AD is substantially horizontal. The rotary damper 23 projects from the inside of the case 21 to the outside of the case 21 through the opening 21a provided in the case 21. More specifically, the outer cylinder 31 of the rotary damper 23 has a first end portion 31a and a second end portion 31b as both end portions of the axial direction AD. The first end portion 31a is located outside the case 21 between the connecting portion Br3 of the Y-shaped brace Br and the outer surface of the case 21. On the other hand, the second end portion 31b is located on the opposite side of the first end portion 31a and is housed inside the case 21.

For example, the rotary damper 23 includes a rotating body 32 inside the outer cylinder 31, and has a damping characteristic due to the rotation of the rotating body 32 with respect to the outer cylinder 31. For example, the rotary damper 23 may be an oil damper, a viscous damper, a viscoelastic damper, or the like.

Next, the fixing mechanism 24, the transmission unit 25, and the actuator 26 will be described. The fixing mechanism 24 fixes the rotary damper 23 to the building BL. The fixing mechanism 24 includes a first fixing portion 51 and a second fixing portion 52.

The first fixing portion 51 is attached to the end portion of the shaft member 36 protruding from the outer cylinder 31. The first fixing portion 51 fixes the end portion of the shaft member 36 to the connecting portion Br3 of the Y-shaped brace Br. In other words, the first fixing portion 51 fixes the position (relative position) of the shaft member 36 in the axial direction AD with respect to the connecting portion Br3 of the Y-shaped brace Br. As a result, when the connecting portion Br3 of the Y-shaped brace Br is displaced in the axial direction AD when vibration is generated, the shaft member 36 of the rotary damper 23 is displaced in the axial direction AD integrally with the connecting portion Br3 of the Y-shaped brace Br. .. For example, the shaft member 36 of the rotary damper 23 is fixed by the first fixing portion 51 so as not to rotate with respect to the connecting portion Br3 of the Y-shaped brace Br.

The second fixing portion 52 is provided between the second end portion 31b of the outer cylinder 31 of the rotary damper 23 and the fixing member 22. The second fixing portion 52 fixes the outer cylinder 31 of the rotary damper 23 to the fixing member 22. In other words, the second fixing portion 52 fixes the outer cylinder 31 of the rotary damper 23 to the leg portion of the second pillar P2 of the building BL, for example, via the fixing member 22. That is, the second fixing portion 52 fixes the position (relative position) of the outer cylinder 31 in the axial direction AD with respect to the leg portion of the second pillar P2. As a result, when the second pillar P2 is displaced in the axial direction AD when vibration is generated, the outer cylinder 31 of the rotary damper 23 is displaced in the axial direction AD integrally with the leg portion of the second pillar P2. However, the configuration of this embodiment is not limited to the above example. For example, the second fixing portion 52 may be fixed to the floor portion F directly or via the fixing member 22 or the case 21. In this case, the outer cylinder 31 of the rotary damper 23 may be displaced integrally with the floor portion F instead of being displaced integrally with the second pillar P2. Further, as another example, the outer cylinder 31 of the rotary damper 23 may be integrally displaced with a member different from the second pillar P2 and the floor portion F.

The second fixing portion 52 of the present embodiment rotatably supports the outer cylinder 31 of the rotary damper 23 with respect to the second pillar P2. More specifically, the second fixing portion 52 has a support shaft 56 and a rotation support mechanism 57.

The support shaft 56 is fixed to the second end 31b of the outer cylinder 31 of the rotary damper 23 and can rotate integrally with the outer cylinder 31. The support shaft 56 extends from the second end 31b of the outer cylinder 31 toward the fixing member 22 along the axial direction AD. The end portion of the support shaft 56 located near the fixing member 22 has a diameter-expanded portion 56a in which the diameter of the support shaft 56 is increased.

The rotation support mechanism 57 is fixed to the fixing member 22. The rotation support mechanism 57 has a bearing 57a and a holder 57b. The bearing 57a rotatably supports the support shaft 56 and the outer cylinder 31 with respect to the fixing member 22 (that is, with respect to the second pillar P2).

With the above configuration, the second fixing portion 52 fixes the position of the axial AD of the outer cylinder 31 with respect to the second pillar P2 (for example, the leg portion of the second pillar P2), and the second pillar P2. The outer cylinder 31 is rotatably supported. In other words, the outer cylinder 31 is rotatably supported by the second fixing portion 52 with respect to the building BL, the case 21, and the actuator 26.

Next, the transmission unit 25 will be described.
The transmission unit 25 is provided between the actuator 26 and the outer cylinder 31 of the rotary damper 23, and transmits the power from the actuator 26 to the outer cylinder 31 of the rotary damper 23. The transmission unit 25 is, for example, a gear provided on the outer peripheral surface of the outer cylinder 31. The transmission portion 25 is formed in an annular shape along the outer peripheral surface of the outer cylinder 31, for example, and is provided over the entire circumference of the outer peripheral surface of the outer cylinder 31. The configuration and mounting position of the transmission unit 25 are not limited as long as the power from the actuator 26 can be transmitted to the outer cylinder 31 of the rotary damper 23. The transmission unit 25 may be formed separately from the outer cylinder 31 and attached to the outer cylinder 31, or may be integrally molded with the outer cylinder 31. Further, the transmission unit 25 is not limited to the gear, and may be a high friction member (for example, a rubber member) or the like.

The actuator 26 is provided outside the rotary damper 23. The actuator 26 is, for example, an electric actuator, and has a motor (drive source) 61, a speed reducer 62, and a gear 63. The speed reducer 62 is connected to the motor 61, and power is transmitted from the motor 61. The gear 63 is connected to the speed reducer 62, and power is transmitted from the speed reducer 62. The gear 63 is engaged with a transmission portion 25 provided in the outer cylinder 31 of the rotary damper 23. The motor 61 causes the transmission unit 25 to act on power (rotational torque) by rotating the gear 63 via the speed reducer 62. The speed reducer 62 is not an essential component and may be omitted. In this case, the motor 61 may be directly connected to the gear 63.

By applying power to the transmission unit 25, the actuator 26 directly applies power (rotational torque) to the outer cylinder 31 of the rotary damper 23 via the transmission unit 25. As a result, the actuator 26 rotates the outer cylinder 31 of the rotary damper 23 with respect to the building BL. In addition, "power is directly applied to the outer cylinder of the rotary damper" means that power is applied to the outer cylinder 31 of the rotary damper 23 without passing through the shaft member 36 of the rotary damper 23 (shaft member 36). It means that the outer cylinder 31 is rotated without the intervention of. In other words, "directly applying power to the outer cylinder of the rotary damper" means that the outer cylinder 31 is operated regardless of the position of the shaft member 36 with respect to the outer cylinder 31 and the moving speed (rotational speed of the rotating body 32). It means to forcibly rotate from the outside.

The actuator 26 rotates the outer cylinder 31 with respect to the building BL, thereby rotating the outer cylinder 31 with respect to the rotating body 32 of the rotary damper 23. As a result, the actuator 26 changes the relative rotation speed between the outer cylinder 31 and the rotating body 32, and changes the magnitude and generation timing of the damping force acting between the outer cylinder 31 and the rotating body 32. , The damping characteristic of the vibration damping device 11 is changed.

For example, the actuator 26 receives an external force input to the shaft member 36 (for example, a force acting on the shaft member 36 due to a relative displacement between the Y-shaped brace Br and the leg portion (or floor portion F) of the second pillar P2). When the shaft member 36 moves with respect to the outer cylinder 31 and the rotating body 32 rotates in the first direction with respect to the outer cylinder 31, the outer cylinder 31 rotates with respect to the rotating body 32 in the second direction opposite to the first direction. It is possible to output the power to make it. The actuator 26 can increase the damping force of the vibration damping device 11 by rotating the outer cylinder 31 in the direction opposite to the rotating body 32 rotating in the first direction.

Further, in the actuator 26, when the shaft member 36 moves by the external force input to the shaft member 36 and the rotating body 32 rotates in the first direction with respect to the outer cylinder 31, the outer cylinder 31 with respect to the rotating body 32. Can output power to rotate in the same direction as the first direction. The actuator 26 can reduce or reduce the damping force of the vibration damping device 11 by rotating the outer cylinder 31 in the same direction with respect to the rotating body 32 rotating in the first direction.

According to the vibration damping device 11 as described above, the damping characteristic of the vibration damping device 11 is dynamically changed by rotating the outer cylinder 31 of the rotary damper 23 by the actuator 26, and the vibration of the building BL is further increased. It can be effectively damped. For example, the vibration damping device 11 changes the rotational state of the outer cylinder 31 of the rotary damper 23 by changing the output of the actuator 26 (rotational speed of the motor 61, etc.) in response to the vibration of the building BL, and the building BL Vibration can be suppressed more effectively.

Next, the outline of the operation of the vibration damping control system 1 of the present embodiment will be described. FIG. 4 is a configuration diagram of a control system of the vibration damping control system 1 of the embodiment. The vibration damping control system 1 implements vibration damping control of a building BL in a real space.

The control unit 13 (controller 13 in FIG. 4) performs feedback control on the building BL as a control target and returns a state value (state s) indicating the state of the building BL based on the vibration data output by the sensor 12. .. The state value to be fed back may be appropriately delayed by a delay element (not shown). The control target in the vibration damping control system 1 is that the building BL does not vibrate even if a disturbance occurs, or that the vibration generated by the disturbance is damped. Seismic motion is an example of a disturbance acting on a building BL.

Such a control unit 13 controls the actuator 26 of the vibration damping device 11 based on the vibration detection result (vibration data) by the sensor 12 to rotate the outer cylinder 31 of the rotary damper 23.

As a result, the vibration damping control system 1 has the shaft of the rotary damper 23 based on the relative displacement between the connecting portion Br3 of the Y-shaped brace Br and the leg portion (or floor portion F) of the second pillar P2 when vibration is generated. In addition to rotating the rotating body 32 by moving the member 36, the actuator 26 additionally rotates the outer cylinder 31 to cause damping acting between the outer cylinder 31 and the rotating body 32 of the rotary damper 23. The magnitude of force and the timing of generation can be adjusted. Further, as a result, the vibration damping control system 1 exerts an active type vibration damping action, and can perform more effective vibration damping than the passive type vibration damping.

The control unit 13 of the embodiment uses a neural network 13NN (first artificial neural network, hereinafter simply referred to as NN) that has been trained using a vibration simulator 110 that simulates the vibration of the building BL to be controlled. Be prepared. The NN of the control unit 13 generates a control amount for the vibration damping device 11 based on a physical quantity representing the movement of each part of the building BL. The physical quantity representing the movement of each part of the building BL is, for example, vibration data of the sensor 12. Such a control unit 13 can control the vibration damping device 11 by the NN based on the vibration data of the sensor 12. The above NN configuration example will be described later.

The control amount for the vibration damping device 11 may be, for example, adjusting any of the shaft length, the shaft expansion / contraction speed, the shaft rigidity, and the shaft force of the rotary damper 23. The above-mentioned shaft length is, for example, the amount of protrusion of the shaft member 36 with respect to the outer cylinder 31 in the rotary damper 23. The shaft expansion / contraction speed is the relative moving speed of the shaft member 36 with respect to the outer cylinder 31 in the moving direction of the shaft member 36. The shaft rigidity is the torsional rigidity of the shaft member 36 around the shaft. The axial force is a resistance force when the shaft member 36 moves with respect to the outer cylinder 31. Based on the control amount for the vibration damping device 11 determined in this way, it is possible to adjust the horizontal couple acting on the two floor portions sandwiching the specific layer in which the vibration damping device 11 is arranged in the building BL. can.

Next, learning of the NN of the control unit 13 will be described. The NN of the control unit 13 of the embodiment acquires the vibration control rule determined by learning before applying it to the actual building BL. As a result, the response characteristics of the control unit 13 are adjusted. The learning of the NN of the control unit 13 is carried out by reinforcement learning by the machine learning device 10. The NN of the control unit 13 does not limit the acquisition of the vibration control rule after being applied to the actual building BL, for example, the readjustment of its response characteristics.

The machine learning device 10 of the embodiment will be described with reference to FIG. FIG. 5 is a block diagram of the machine learning device 10 of the embodiment.

The machine learning device 10 (vibration analysis device) includes a vibration analysis unit 100, a reward generation unit 150, and a learning control unit 170.

The vibration analysis unit 100 includes a vibration simulator 110 and a controller 130.

The vibration simulator 110 simulates the vibration of a virtual building (virtual building structure model) that replaces the actual building BL. The vibration simulator 110 of the embodiment includes a virtual actuator 111, a virtual sensor 112, and a virtual building body 113. The virtual actuator 111 and the virtual sensor 112 may correspond to the vibration damping device 11 and the sensor 12, respectively. The virtual building body 113 mainly corresponds to the structural members of the building BL.

The virtual actuator 111 exerts stress on the virtual building body 113 in response to control. The stress acting on the virtual building body 113 includes a case where a couple acts on at least two floor portions sandwiching a specific layer. The vibration simulator 110 reproduces the vibration generated in the virtual building main body 113 when the disturbance based on the predetermined vibration data and the stress by the virtual actuator 111 act on the virtual building main body 113.

The virtual sensor 112 detects vibration at a predetermined position on the virtual building body 113. For example, vibration is detected as a physical quantity having a magnitude and direction such as displacement, velocity, and acceleration, and is recorded as time-series vibration data.

The vibration simulator 110 carries out a vibration simulation using the virtual building structure model shown in FIG. FIG. 6 is a diagram for explaining a virtual building structure model of the embodiment. FIG. 6A shows a multi-mass model of a building BL built on the ground. M1 to Mm indicate the mass points of each layer from the first layer to the mth layer. Seismic motion as a disturbance is transmitted from the ground side to the bottom layer (foundation) of the building BL.

FIG. 6B shows the multi-mass model of FIG. 6A as a vibration model. According to FIG. 6, the vibration damping device 11 is provided between the bottom layer and the first layer, between the first layer and the second layer, and is provided in the other layers (interlayers). Indicates a non-existent state. In this case, there are elements k1, c1, y1 between the bottom layer and the first layer, elements k2, c2, y2 between the first layer and the second layer, and the second layer and the third layer. There is an element k3 between. Although the description is omitted, the third layer and above also have elements k4 to km. Elements k1 to km are spring constants. The elements c1 and c2 are damping coefficients. Elements y1 and y2 are quantities controlled by the actuator.

The vibration simulator 110 of the embodiment reproduces the vibration of the building BL by using the above-mentioned multi-mass model by the vibration simulation.

Returning to FIG. 5, the controller 130 will be described. The controller 130 is a model of the control unit 13 shown in FIG. 1, and includes an NN 131 (second artificial neural network) that models the same configuration as the NN of the control unit 13.

By connecting the vibration simulator 110 and the controller 130, the vibration analysis unit 100 implements vibration damping control of the virtual building structure model by feedback control in the same manner as the control system of the actual vibration damping control system 1. ..

The reward generation unit 150 generates a reward r based on the analysis result by the vibration analysis unit 100.
The reward r is used for reinforcement learning. The reward generation unit 150 of the embodiment generates a reward r based on the magnitude (state s) of vibration of the virtual building structure model when controlled by the controller 130. For example, the function R (・) for calculating the reward r is shown in the equation (1).

r = R (s) = R (x x ″ x ″) ・ ・ ・ (1)

In equation (1), s is a vector indicating the state. x, x ″, and x ″ are vectors indicating the amount of vibration and the displacement, velocity, and acceleration of each layer. The function R takes the state s as a variable, or takes some or all of the displacement, velocity, and acceleration of each layer as variables. Such a function R can be defined as a linear or non-linear arithmetic expression. The function R may be composed of a combination of a plurality of arithmetic expressions. The above arithmetic expression may include a determinant, a difference equation, and the like.

For example, a function R is defined in which the absolute values of x, x ′, and x ″ are added and multiplied by a negative coefficient. This function R (x x ″ x ″) indicates a larger value if the value of each variable is smaller. The negative value of the function R increases as the amount of vibration of the building BL increases or the vibration becomes more intense. The reward generation unit 150 calculates a reward r having a value corresponding to the magnitude of vibration. From the value of the reward r, it is possible to identify that the building BL has learned to a more preferable state by identifying a state in which the vibration amount of the building BL is small or a state in which the vibration is gentle.

The function R is not limited to this, and each coefficient and other forms of arithmetic expressions may be set as appropriate. The reward generation unit 150 calculates the reward r using the above formula (1).

The learning control unit 170 updates the action policy or action value corresponding to the current state variable (state s) and the possible action (action a) based on the update formula and the reward r described later.

For example, the vibration analysis unit 100 receives an instruction of an action policy from the learning control unit 170. The action policy in the embodiment includes the characteristic information of NN131, the type of vibration waveform applied to the simulation, and the like. The characteristic information of NN131 includes, for example, a weight W, which will be described later. The characteristic information of NN131 is optimized by reinforcement learning (deep reinforcement learning).

Each configuration of the machine learning device 10 shown in FIG. 5 may be grasped as follows. The vibration simulator 110 and the reward generation unit 150 are collectively referred to as "environment 110E". The learning control unit 170 and the controller 130 are collectively referred to as an "agent 170A". The machine learning device 10 including the "environment 110E" and the "agent 170A" can be treated as a form of so-called reinforcement learning.

Next, a case where the Q-learning method and the ε-greedy method are applied will be described as an example of the reinforcement learning algorithm applied to the embodiment.

The Q-learning method is called a value function Q (sk, ak) (hereinafter referred to as a function Q (s, a)) for selecting an action a under a certain environmental state s (hereinafter, simply referred to as a state s). The value is called the Q value.) This is a method of learning. Hereinafter, the value function Q (sk, ak) is simply referred to as a function Q (s, a).

For example, in a certain state s, it is required to select the action a having the highest value as the optimum action. The value when the action a is selected in such a state s is indicated by the Q value obtained by the function Q (s, a).

Even if a function Q (s, a) that cannot be approximated to the extent that the correct action a can be selected is used, the correct Q value cannot be obtained. Therefore, the learning control unit 170 enhances the eligibility of the function Q by performing learning that gives a reward r that the better action a is selected. The learning control unit 170 learns the function Q (s, a) that can select a better action a by repeating the trial of selecting the action a under a certain state s.

From the result of the action, we derive a function Q (s, a) optimized for the goal of increasing (maximizing) the total amount of rewards that will be obtained in the future rather than obtaining short-term value. do. The update formula of the function Q (s, a) is shown in equation (2).

Here, sk represents the state of the environment at time k, and ak represents the action at time k. The state changes to sk + 1 by the action ak. rk + 1 represents the reward that can be obtained by changing the state. In addition, the term with max is the highest Q value multiplied by γ in the trial of the action a that can be taken under the state sk + 1. γ is a discount rate and takes a value in the range of 0 <γ ≦ 1. α is a learning coefficient and takes a value in the range of 0 <α ≦ 1.

This equation (2) represents an example of updating Q (sk, ak), which is the evaluation value of the action ak in the state sk, based on the reward rk + 1 obtained as a result of the action ak.

By reinforcement learning using the function Q (s, a) as described above, it is possible to obtain the NN of the desired vibration control rule (response characteristic) having a high Q value, that is, an evaluation value.

Next, the outline of the ε-greedy method will be described.
When learning progresses to some extent, the action a that maximizes the value of the function Q (s, a) becomes close to the action that should actually be selected. However, depending on the conditions, the desired learning may not proceed.

Therefore, when selecting the action a, the controller 130 not only selects the action a that maximizes the value of the function Q (s, a), but also performs a random action at a certain ratio ε at a ratio (1-ε). The action that maximizes the value of the function Q (s, a) is selected (ε-greedy method). As a result, it is possible to try the action that maximizes the function Q (s, a) in the process of calculation, including that the action is not always the optimum action.

The process for optimizing the control unit 13 will be described later.

Next, the NN of the embodiment will be described with reference to FIGS. 7 and 8. FIG. 7 is a schematic diagram showing a model of neurons of the embodiment. The NN is composed of an arithmetic unit and a memory that imitate a neuron model as shown in FIG. 7, for example.

As shown in FIG. 7, the neuron outputs an output y for a plurality of inputs x (here, as an example, inputs x ₁ to input x _3). Each input _x 1 ~x _3, the weight _w corresponding to the input x _(w 1 ~w ₃₎ is multiplied. As a result, the neuron outputs the output y expressed by the following equation. The input x, the output y, and the weight w are all vectors. The above relationship is shown in equation (3).

In the above equation (3), θ is the bias and f _k is the activation function. As the activation function, one that imitates a sigmoid curve, one that shows a quantized discrete value, or the like may be applied.

Next, an example of the NN in which the above-mentioned neurons are combined will be described with reference to FIG. FIG. 8 is a schematic view showing the NN of the embodiment. The NN shown in FIG. 8 has a three-layer structure of D1 to D3. D1 is an input layer, D2 is an intermediate layer, and D3 is an output layer.

As shown in FIG. 8, a plurality of inputs x (here, as an example, inputs x1 to x3) are input from the left side of the NN, and a result y (here, as an example, result y1 and result y2) is output from the right side. NS.

Specifically, in the input layer D1, the three neurons N11 to N13 multiply each of the inputs x1 to the input x3 by the corresponding weights to take the sum, and output Z11 to Z13 generated by the activation function, respectively. do. These weights are collectively labeled W1. These Z11 to Z13 are collectively referred to as a feature vector Z1. The feature vector Z1 indicates a feature quantity derived from an input vector having inputs x1 to input x3 as elements.

In the intermediate layer D2, two neurons N21 and N22 multiply each of Z11 to Z13, which are elements of the feature vector Z1, by corresponding weights to sum, and output Z21 and Z22 generated by the activation function, respectively. do. These weights are collectively labeled W2. These Z21 and Z22 are collectively referred to as a feature vector Z2. The feature vector Z2 indicates a feature quantity derived from the feature vector Z1.

In the output layer D3, the two neurons N31 and N32 multiply each of the elements Z21 and Z22 of the feature vector Z2 by the corresponding weights and sum, and the result y1 and the result y2 generated by the activation function are obtained. Output each. These weights are collectively labeled W3. The result y1 and the result y2 are derived from the feature vector Z1.

At the learning stage, the NN configuration and weights W1 to W3 are determined. The determined NN configurations and weights W1 to W3 are applied to the NN of the control unit 13 of the vibration damping control system 1 of the actual building BL. The number of such NN layers can be increased to three or more by making a plurality of intermediate layers.

The NN 131 of the controller 130 includes, for example, the NN as described above. The NN 131 adjusts the weight W of the NN 131 in association with the instruction of the action policy by the learning control unit 170. The NN 131 calculates the action a using the weight W corresponding to the action policy under the state s based on the measurement data of the virtual sensor 112. That is, the weight W is associated with the function (S, a). For example, the weight W may be stored in the storage area in the form of a look-up table. The calculated action a is a control command for the virtual actuator 111.

After adjusting the NN 131 according to the instruction of the action policy, the controller 130 sequentially calculates the control command value for the virtual actuator 111 under the state s based on the measurement data of the virtual sensor 112. The controller 130 may use either the result y1 or the result y2 of the NN as the control command value, or may use the value derived by combining the result y1 and the result y2 of the NN as the control command value. For example, the controller 130 sets the maximum value among the result y1 and the result y2 of the NN as the control command value.

In the vibration analysis unit 100, the controller 130 constitutes a feedback control system in which the state s output by the vibration simulator 110 is fed back. The control goal of this control system is to prevent the virtual building structure model from vibrating even if a disturbance occurs, or to attenuate the vibration generated by the disturbance.

The vibration analysis unit 100 connects the vibration simulator 110 and the controller 130 to perform a vibration analysis simulation of the virtual building structure model, and as a result, outputs a state s based on the measurement data of the virtual sensor 112. ..

Next, the operation of the machine learning device 10 according to the embodiment of the present invention will be described.

In the process of adjusting the control unit 13, the machine learning device 10 of the embodiment vibrates by numerical analysis using a virtual building structure model of the building BL instead of vibration analysis (reinforcement learning) using the actual building BL. Conduct simulation (reinforcement learning).

As a result, the vibration damping control system 1 enables the following. For example, even if it is difficult to use the actual building BL in order to optimize the control unit 13, the vibration damping control system 1 can be constructed by using the vibration simulation. In addition, seismic waves rarely arrive at the position of the building BL, and there are not many enough to efficiently carry out reinforcement learning, but by using vibration simulation, sufficient learning can be carried out efficiently in a short period of time. As a result, an appropriate damping effect can be obtained for an earthquake that actually occurs from the initial stage when the damping control system 1 is applied to the actual building BL. Furthermore, when the target building BL does not actually exist at the design stage of the building BL, it is possible to analyze the building BL scheduled in the future by using the vibration simulation.

In the machine learning device 10 of the embodiment, the vibration analysis unit 100 uses a vibration simulation based on a virtual building structure model that models the building BL and vibration data that corresponds to disturbances to the building BL (hereinafter referred to as disturbance data). Simulates the vibration of each part in the virtual building structure model. As a result, the vibration analysis unit 100 generates data to be used for the adjustment of the control unit 13.

For example, when the vibration caused by the disturbance data is transmitted to the virtual building, the virtual building structure model is shaken. The vibration analysis unit 100 reproduces the behavior of the virtual building structure model described above. Further, the vibration analysis unit 100 reproduces the stress applied to the virtual building structure model due to the vibration indicated by the disturbance data and the stress applied to the virtual building structure model by the virtual actuator 111. The vibration analysis unit 100 outputs vibration data related to vibration of each unit instead of the actual sensor 12.

The machine learning device 10 of the embodiment carries out the reinforcement learning process for optimizing the control unit 13 according to the following procedure. FIG. 9 is a flowchart showing the procedure of the reinforcement learning process of the embodiment.

Step SA11:
The vibration analysis unit 100 acquires information on the virtual building structure model of the building BL and the characteristics and arrangement positions of the vibration damping device 11.

Step SA12:
The vibration analysis unit 100 determines the initial values of the state s, the reward r, and the action a for reinforcement learning.

Step SA13:
The vibration analysis unit 100 determines the disturbance data to be applied to the vibration simulation. For example, the disturbance data includes all waveforms applied to the vibration simulation for one episode.

Step SA14:
The learning control unit 170 instructs the vibration analysis unit 100 to take action measures, and causes the vibration analysis unit 100 to perform a vibration simulation based on the disturbance data.

Step SA15:
Vibration analysis unit 100 in accordance with the above action policy, determines the action a _{k + 1} based on the state s _k in episodes, implemented vibration simulations based on disturbance data, sequentially records the state s _{k + 1} is the result do. The process of step SA15 is carried out sequentially from the beginning to the end of the episode.

Step SA16:
The reward generation unit 150 derives the _{reward r k + 1} based on the _{state sk + 1} , which is the result of the vibration simulation for one episode. The derivation of the reward r _{k + 1} may be performed after the processing of step SA15 is completed, or may be performed in parallel with the processing of step SA15.

Step SA17:
The learning control unit 170 _{determines a Q value corresponding to the next action policy based on the action ak} , the state _{sk + 1,} and the reward rk _{+ 1} , and notifies the vibration analysis unit 100 of the Q value.

The learning control unit 170 repeats the processes of steps SA13 to SA17 as the analysis process of the next episode, and executes the analysis process of the scheduled number of episodes.

As the seismic waveform as the disturbance data used for the above reinforcement learning (machine learning), the observation record observed at the site of the building BL may be used, or the observation record observed at a point away from the building BL or the observation record. Observation records of past waves may be used. Further, the acceleration waveform calculated by calculation instead of observation may be used.

According to the above embodiment, the control unit 13 of the vibration damping control system 1 includes an NN that has undergone learning processing using a vibration simulator 110 that simulates the vibration of a building BL. The control unit 13 has a simpler method by adjusting the force that the stress adjusting unit acts as a horizontal couple on the two floor portions sandwiching the specific layer in the building BL based on the output of the NN. , The vibration of the building to be controlled can be suppressed. For example, the vibration simulator 110 of the embodiment increases the amount of deformation of the specific layer by applying a couple in the horizontal direction to two floor portions sandwiching the specific layer of the virtual building structure model of the building BL. Therefore, the effect of vibration damping control on the actual building BL can be easily verified without using the actual building BL.

Further, the vibration damping control system 1 is set in the vibration damping control system 1 by acquiring a vibration control model of the control system applicable to the actual building BL by reinforcement learning (deep reinforcement learning) using the vibration simulator 110. The vibration suppression performance can be improved, and the accuracy for a desired control state can be improved.

Further, if the movement (sway) of the building BL to be controlled can be detected by the sensor 12, the NN of the control unit 13 controls the vibration damping device 11 based on the physical quantity representing the actual movement of the building BL. Can be done. The position of the sensor 12 in this case may be aligned with the position where the virtual sensor 112 arranged in the virtual building structure model of the building BL is arranged. The physical quantity representing the movement of the building BL may be any one related to the deformation of the layer, such as the inter-story deformation angle (interlayer displacement), the absolute displacement of the layer, the velocity, and the acceleration. Further, as the physical quantity, the maximum value or the average value of the physical quantity detected at a plurality of positions in each floor may be appropriately defined.

The vibration damping device 11 can suppress the vibration of the building BL by applying a couple adjusted by the above control to the two floor portions sandwiching the specific layer of the building BL.

Further, according to the embodiment, the state value (state s) is generated by performing the response analysis. The learning control unit 170 determines the feature information of the NN 131 by performing reinforcement learning with the degree (degree) of vibration control based on the state value (state s) as the reward r. As a result, the feature information of the NN of the control unit 13 is determined.

The control unit 13 determines the NN feature information in advance by reinforcement learning, and may store a plurality of feature information having different features in a predetermined storage area as the feature information. By defining the control model corresponding to each of the plurality of feature information in this way, the optimum control method for the NN of the control unit 13 can be instantly selected and applied.

The vibration damping control system 1 acquires a vibration control model of a control system with high performance and accuracy by reinforcement learning (deep reinforcement learning) using the vibration simulator 110.

For example, the disturbance data input to the virtual building structure model is acceleration data (ground motion acceleration) including a part or all of the components in two horizontal directions and one vertical direction, and shows a time history waveform. It is a thing.

(First modification of the first embodiment)
The first modification is an example in which a reinforcement learning algorithm different from that of the first embodiment is applied. The control unit 13 of this modified example has its response characteristics adjusted as the control unit 13 based on the vibration control rule determined by reinforcement learning. The case where the policy gradient method is applied to reinforcement learning will be described below.

The policy gradient method is a learning method that enables a more suitable action to be selected by optimizing the action policy for selecting the action a under a certain environmental state s. The action policy is defined by the policy function (hereinafter referred to as the function P (s, a)) shown in the equation (4).

For example, the learning control unit 170 learns the function P (s, a) based on the probability or the number of actions a taken in a certain state s. For example, the function P (s, a) shown in the equation (4) is updated based on the number of times N1 of taking the action a in the state s and the number of times N2 of not taking the action a.

In the above equation (4), α1 and α2 are predetermined constants. The values of α1 and α2 are set so that the action a is easily selected when the reward r is equal to or more than the expected value, and the action a is difficult to be selected when the reward r does not satisfy the expected value.

The learning control unit 170 updates the function P (s, a) of the above equation (4) every time the evaluation of a predetermined episode is performed based on the policy. An episode in an embodiment is a processing unit that performs one trial using all or part of a predetermined type of disturbance applied to a building BL. At that time, the learning control unit 170 sets "+ α1" when the magnitude of the vibration is equal to or less than the predetermined value, and the magnitude of the vibration exceeds the predetermined value, based on the reward r based on the result of the trial for each episode. In this case, "-α2" is added to the function P (s, a).

The function P (s, a) is learned by randomly generating an action a'different from the action a recommended based on the above equation (4) so that the learning result does not stay in the locally optimal solution. May proceed.

By appropriately updating the function P (s, a) according to the above equation (4), a function that leads to a policy that the action a should be executed in a certain state s is derived.

The function P (s, a) derived here is determined to have a larger value when _{a suitable action ai} is selected from the actions a that can be taken under the state s.

The learning control unit 170 carries out learning about the behavioral policy regarding the NN 131 according to the above learning method. As described above, the vibration control rule determined by reinforcement learning using the function P (s, a) becomes a desired one, so that the NN of the control unit 13 of the same vibration control rule is desired. It will have response characteristics.

Although the efficiency of learning and the accuracy of learning results are improved by using reinforcement learning as shown in the first embodiment and the first modification, the above-mentioned reinforcement learning method is used. However, a learning method for known NNs may be applied.

(Second Embodiment)
In the second embodiment, a more specific example of the vibration simulation by the vibration analysis unit 100 will be shown. The vibration analysis unit 100 drives the virtual building structure model and performs response analysis to generate state values representing the movements of each part of the virtual building structure model.

Since the vibration analysis unit 100 generates an analysis model as a virtual building structure model, the basic model (formula (5) described later) for generating this analysis model and the movement of the ground and the building to be analyzed in the design documents and the like. The design data for calculating the characteristic matrix (matrix of mass, attenuation, and rigidity, which will be described later) is read out from the storage area. This design data includes, for example, physical property values (density, elastic wave velocity, damping constant, etc.) for each layer of the ground, and constants (mass, rigidity, damping constant, etc.) for each floor indicating the vibration characteristics of the building.

In addition, the analysis model can be a mass point system model or a three-dimensional skeleton model. When this mass point system model is used, the constants (mass, rigidity, damping constant, etc.) of each floor can be obtained from the three-dimensional skeleton elasto-plastic analysis model (three-dimensional frame model) of the building.
As an example of the mass point system model, there is a shear multi-mass point system model. The seismic response of the shear multi-mass model can be obtained by solving the equation of motion shown in the following equation (5).

[M0] {x ″} ＋ [C0] {x ″} ＋ [K0] {x} ＝ {P} ・ ・ ・ (5)

In this equation (5), [M0] is a mass matrix, [C0] is a damping matrix, and [K0] is a stiffness matrix. Further, each matrix element (hereinafter, simply referred to as an element) of the mass matrix [M0], the damping matrix [C0], and the rigidity matrix [K0] has the default values of the number of columns and the number of rows of the elements. Generally, the external force {P} in the seismic response analysis is a force (inertial force) acting on each mass point due to the seismic motion (ground motion acceleration).

Further, in this equation (5), x ″ indicates the relative acceleration of the mass point (acceleration in the direction parallel to the ground surface), x ′ indicates the relative velocity, and x indicates the relative displacement. {X''} is a column vector (m × 1 type matrix) showing the acceleration of each mass point in the parallel direction on the ground surface of the mass point, which is the analysis position (evaluation target position) in the direction perpendicular to the ground surface. .. As shown below, each of the acceleration {x ″}, velocity {x ″}, displacement {x} and external force {P} in the equation (5) is a column vector shown in the following equation (6). In the following formula, the number of floors of the building is set to m floor for convenience.

{x''} = [x1''x2''x3'' ... xm''] ^T
{x'} = [x1'x2'x3'... xm'] ^T
{x} = [x1 x2 x3 ... xm] ^T
{P} = [P1 P2 P3 ... Pm] ^T
... (6)

That is, the relative acceleration {x''}, which is the column vector in the equation (6), is relative to the ground surface at each mass point (for example, corresponding to the number of floors of the building), which is the analysis position in the direction perpendicular to the ground surface. It is a column vector showing acceleration in parallel directions. The relative velocity {x'} is a column vector indicating the velocity in the direction parallel to the ground surface at each mass point which is the analysis position in the direction perpendicular to the ground surface. The relative displacement {x} is a column vector indicating the displacement in the direction parallel to the ground surface at each mass point which is the analysis position in the direction perpendicular to the ground surface. The external force {P} is a column vector indicating the inertial force acting in the direction parallel to the ground surface at each mass point which is the analysis position in the direction perpendicular to the ground surface.

The vibration analysis unit 100 extracts the matrix elements of the default dimensions of the mass matrix [M0], the damping matrix [C0], and the rigidity matrix [K0] in the equation (6) from the floor number of the building BL and the above design data, respectively. Change to a matrix element of the dimension calculated by the finite element method or the like.

Further, the elements of each matrix are calculated in advance from the design data, and are written and stored in the database 16 in advance corresponding to the number of floors of the building BL and the design data. Then, the vibration analysis unit 100 reads out the elements of each matrix corresponding to the building BL from the database 16, expands and modifies the basic model according to the dimensions of each element of the matrix, and generates an analysis model of the building. You may do so.

Then, the vibration analysis unit 100 changes (extends and changes) the dimensions of each element of the mass matrix [M0], the damping matrix [C0], and the rigidity matrix [K0] in the basic model according to the floor number of the building BL. By doing so, the mass matrix [MD], the damping matrix [CD], and the rigidity matrix [KD] are obtained, and the column vectors of acceleration {x''}, velocity {x'}, and displacement {x} are vectorized according to the rank. Expand the number of elements in. Then, the vibration analysis unit 100 includes the obtained mass matrix [MD], damping matrix [CD], and stiffness matrix [KD], and a column vector of acceleration {x''}, velocity {x'}, and displacement {x}. And, the equation (6) is changed to the following equation (7) to generate an analysis model of the building BL to be analyzed.

[MD] {x''} + [CD] {x'} + [KD] {x} =-[MD] {1} y0''
... (7)

In this equation (7), x ″ indicates the relative acceleration of the mass point (acceleration in the direction parallel to the ground surface), x ′ indicates the relative velocity, and x indicates the relative displacement. Further, y0 ″ indicates the acceleration due to the seismic motion (ground motion acceleration), and the right side of the equation (7) explicitly expresses the inertial force acting on each mass point due to the seismic motion as an external force. The acceleration {x ″}, the velocity {x ″}, and the displacement {x} are configured as response values with each floor as a mass point, as described above.

Here, a couple of the magnitude of f1 is applied to the first layer and f2 is applied to the second layer. At this time, a pair of forces (+ f1 and −f1) having opposite directions act on the upper and lower floors (foundation and mass point 1) of the first layer at the same time. Similarly, a pair of forces (+ f2 and −f2) having opposite directions act on the upper and lower floors (mass point 1 and mass point 2) of the second layer at the same time. Assuming that the direction of the force acting on the upper floor side is positive (+), the force {F} acting on each mass point is expressed by the following equation (8).

{F} = [F1 F2 F3 F4 ... Fm] ^T
F1 = f1-f2, F2 = f2, F3 = F4 = ... = Fm = 0
... (8)

The couple applied between the layers can be considered by applying it to the right side of the equation of motion (5) as an equivalent external force {F} acting on each mass point.

[MD] {x''} + [CD] {x'} + [KD] {x} = {P} + {F}
... (9)

By applying the force (couple) that the virtual actuator 111 acts on between layers as the external force term {F} in the above equation of motion (9), the control force by the virtual actuator 111 is taken into consideration in the seismic response analysis of the mass system model. can do.

Then, the vibration analysis unit 100 writes and stores the generated analysis model of the equation (9) in a storage unit (not shown) in correspondence with the building BL (corresponding to the identification information that identifies the building BL).

The vibration simulator 110 reads out the equation (9) corresponding to the building BL to be analyzed from the storage unit, and arranges it between the ground motion acceleration y0'' due to the seismic motion assumed for this equation (9) and the building. Using the external force {F} acted by the virtual actuator 111, the response values of acceleration {x''}, velocity {x'}, and displacement {x}, which are column vectors in each floor as mass points, are obtained ( Perform response analysis).

That is, the vibration simulator 110 uses a column vector by substituting the inertial force {P} acting by the seismic motion and the force {F} acting by the virtual actuator 111 into the equation of motion of the equation (9) for each fine time interval Δt. Calculate each of a certain acceleration {x''}, velocity {x'}, and displacement {x}.

The inertial force {P} acted by the seismic motion and the force {F} acted by the virtual actuator 111 may be defined by forced displacement using the rigidity matrix [KD].

According to the above embodiment, the vibration damping device 11 enables vibration simulation that reproduces the force that increases the strain of a specific layer of the building BL.

According to the above embodiment, in addition to having the same effect as that of the first embodiment, a response that drives the virtual building structure model to generate a state value representing the movement of each part of the virtual building structure model. The analysis can be carried out.

Although the embodiment of the present invention has been described above, the vibration damping control system 1 and the machine learning device 10 have a computer system inside. A series of processing processes related to the above-mentioned processing is stored in a computer-readable storage medium in the form of a program, and the above processing is performed by the computer reading and executing this program. Here, the computer-readable storage medium refers to a magnetic disk, a magneto-optical disk, a CD-ROM, a DVD-ROM, a semiconductor memory, or the like. Further, this computer program may be distributed to a computer via a communication line, and the computer receiving the distribution may execute the program. Further, the term "computer system" as used herein includes an OS and the like.

Although one embodiment of the present invention has been described above, the embodiment of the present invention is not limited to the above. For example, the methods illustrated in each embodiment and its modifications may be combined in a combination other than the illustrated combinations. Further, the embodiment of the present invention can be a modification of the above embodiment as follows.

For example, in the above embodiment, the configuration related to the present invention has been described by dividing the machine learning device 10 into a vibration analysis unit 100, a reward generation unit 150, and a learning control unit 170 for convenience. The division of the machine learning device 10 may be changed from that illustrated above, or the functional units may be integrated with each other. Moreover, a part of each functional part may be included in another functional part. Further, a part of the machine learning device 10 may be used as a vibration analysis device, and for example, the vibration simulator 110 may be an example of the vibration analysis device.

Further, the control unit 13 may be a computer system, or a part or all of the control unit 13 may be a hardware function unit.

Further, the rotary damper 23 has a vibration damping function of converting the vibration energy of the building BL into heat to apply a damping force to the building BL when vibration occurs, and a shaft member from the outer cylinder 31 of the rotary damper 23. It also has a shaft length adjusting function that actively adjusts the protrusion amount of the 36. Instead of this, the function of the rotary damper 23 may be distributed to a damper having a vibration damping function and an actuator having a shaft length adjusting function. In this case, by arranging the damper and the actuator in series, the actuator can adjust the shaft length of the shaft member of the damper instead.

1 Vibration damping control system 1, 10 Machine learning device (vibration analyzer), 11 Vibration damping device, 12 Sensors, 13 Control unit (controller), 13NN neural network (first artificial neural network), 100 Vibration analysis unit, 110 Vibration Simulator, 111 virtual actuator (virtual stress adjustment unit), 130 controller, 131 NN (second artificial neural network), 150 reward generation unit, 170 learning control unit, 110E environment, 170A agent

Claims (7)

A specific layer among the plurality of layers includes a first artificial neural network (hereinafter referred to as a first NN) that has been trained using a vibration simulator that simulates the vibration of a building structure having a plurality of layers. A vibration damping control system including a control unit that adjusts the force that the stress adjustment unit acts as a couple in the horizontal direction on the two floor portions sandwiched by the first NN. A detector that detects the movement of the building structure,
Based on the control, the stress adjusting part that applies the couple in the building structure to the two floor parts,
With
The first NN of the control unit is
The stress adjusting unit is controlled based on a physical quantity representing the movement of the building structure.
The vibration damping control system according to claim 1. The second artificial neural net of the vibration analysis device that analyzes the vibration based on the virtual building structure model by the response analysis that drives the virtual building structure model of the building structure replaces the stress adjusting unit. And based on the generated control amount, between the first and second quality points of the virtual building structure model corresponding to the two floor portions, or with the first quality point. The virtual stress adjusting unit applies a horizontal even force to the reference plane of the virtual building structure model to perform response analysis for driving the virtual building structure model, and the virtual building structure is driven. State values are generated to represent the movement of each part of the model.
The feature information of the first NN of the control unit is determined by learning using the degree of vibration control based on the state value generated by the vibration simulator as a reward.
The vibration damping control system according to claim 1 or 2. A vibration analysis device that simulates the vibration of a building structure with multiple layers.
Based on the control amount generated by the artificial neural net, the virtual building of the building structure is such that the stress adjusting unit applies a horizontal couple to the two floor portions sandwiching the specific layer among the plurality of layers. A vibration analysis device including a vibration simulator that drives a structure model and generates state values representing the movements of each part of the virtual building structure model by vibration analysis. The vibration simulator generates the state value by dynamically adjusting the couple acting on the two floor portions in the building structure.
The vibration analysis device according to claim 4. An artificial neural net that has been trained using a simulator that simulates the vibration of a building structure having multiple layers causes a stress adjustment unit to act on two floors that sandwich a specific layer among the multiple layers. A vibration damping control method that includes the process of adjusting the couple in the horizontal direction. It is a vibration analysis method of a vibration analysis device that simulates the vibration of a building structure having multiple layers.
Based on the control amount generated by the artificial neural net, the virtual building of the building structure is such that the stress adjusting unit exerts a horizontal couple on two floor portions sandwiching a specific layer among the plurality of layers. A vibration analysis method including a process of driving a structure model and performing response analysis to generate state values representing the movements of each part of the virtual building structure model.

Images