CN-121992891-A - Self-adaptive viscous damping wall based on machine learning

CN121992891ACN 121992891 ACN121992891 ACN 121992891ACN-121992891-A

Abstract

The invention belongs to the technical field of damping walls, in particular to a self-adaptive viscous damping wall based on machine learning, which comprises a channel steel box internally loaded with viscous liquid, wherein the channel steel box is arranged between two damper connecting wall piers, a damping body is vertically arranged in the channel steel box, a self-adaptive mechanism is arranged in the channel steel box in a penetrating manner, and two ends of the self-adaptive mechanism are provided with tackifying mechanisms. The self-adaptive viscous damping wall based on machine learning converts sliding friction into rolling friction through a ball structure, so that movement resistance and abrasion are remarkably reduced, movement smoothness and self-balancing capacity of a damping body under complex multidimensional vibration are guaranteed, additional bending moment and stress concentration caused by mismatching of movement directions and structural rigidity are effectively avoided, meanwhile, a built-in belleville spring buffer module can absorb huge impact energy when extremely overloaded, mechanical hard limit is provided by matching with a limit groove, and double protection is provided for the damping wall.

Inventors

XIN LI
DONG LIGUO
SHI SHENGZHI
JING JING
NI BAOJUN
HOU YUANBO
XU ZHAODONG
ZHU LIHUA
DONG YAORONG
WANG LIANYING

Assignees

中国建筑西北设计研究院有限公司
中建震安科技工程有限公司
西安建筑科技大学

Dates

Publication Date: 20260508
Application Date: 20251214

Claims (10)

1. The self-adaptive viscous damping wall based on machine learning comprises a channel steel box (1) internally loaded with viscous liquid, and is characterized in that the channel steel box (1) is arranged between two damper connecting wall piers (2), a damping body (3) is vertically arranged in the channel steel box (1), the upper surface of the damping body (3) is fixedly connected with the lower surface of the damper connecting wall piers (2) through an embedded part, a self-adaptive mechanism (4) is arranged in the channel steel box penetrating through the damping body (3), and tackifying mechanisms (6) are arranged at two ends of the self-adaptive mechanism (4); The self-adaptive mechanism (4) keeps self-balance on the damping body (3) in the moving process, wherein the damping body (3) reciprocates along the length direction of the damper connecting wall pier (2) through viscous liquid; The tackifying mechanism (6) is fully contacted with viscous liquid in the channel steel box (1) and is bonded and damped during vibration.
2. The machine learning-based self-adaptive viscous damping wall is characterized in that lifting hydraulic cylinders (11) are symmetrically and hingedly mounted on the lower surface of the channel steel box (1), piston rod ends of the lifting hydraulic cylinders (11) are hingedly mounted on the upper surface of the lower damper connecting wall pier (2) through another embedded part, a plurality of buffer telescopic rods (12) for lifting the channel steel box (1) are fixedly mounted on the lower surface of the channel steel box (1) through other groups of embedded parts, and the lower surfaces of the buffer telescopic rods (12) are fixedly mounted on the upper surface of the lower damper connecting wall pier (2) through corresponding embedded parts.
3. The machine learning-based adaptive viscous damping wall of claim 2, wherein the damping body (3) is composed of a transverse connecting top plate (31) and a vertical viscous plate (32) fixedly connected to the lower surface of the connecting top plate (31), the upper surface of the connecting top plate (31) is fixedly connected with the lower surface of the upper damper connecting wall pier (2) through an embedded part, and flow gaps exist between the peripheral side surface and the lower surface of the viscous plate (32) and the inner wall of the channel steel box (1).
4. The machine learning-based self-adaptive viscous damping wall according to claim 3, wherein the self-adaptive mechanism (4) comprises through holes (41) which are formed in a rectangular array on the surface of the viscous plate (32) in a penetrating manner, an outer sleeve (42) is fixedly sleeved on the inner surface of the through holes (41), elliptical grooves (43) are formed in the inner surface of the outer sleeve (42) in a ring-shaped array manner, an inner sphere (44) with an inner hole is arranged in the outer sleeve (42), and supporting shafts (45) penetrating through two sides of the viscous plate (32) are fixedly connected to the inner surface of the inner sphere (44).
5. The machine learning-based adaptive viscous damping wall according to claim 4, wherein the adaptive mechanism (4) further comprises a ball cage (46) slidably connected to the inner surface of the outer sleeve (42), the inner surface of the ball cage (46) is slidably contacted with the outer surface of the inner sphere (44), grooves (47) corresponding to the elliptical grooves (43) one by one are formed in the outer surface of the inner sphere (44) in an annular array distribution, a plurality of window holes (48) uniformly distributed along the circumferential direction are formed in the outer surface of the ball cage (46) in a penetrating mode, each window hole (48) accommodates one ball (49) in a rolling mode, the elliptical grooves (43) are opposite to the grooves (47) and squeeze the balls (49), and accordingly the balls (49) are in rolling contact with the inner surfaces of the elliptical grooves (43) and the grooves (47).
6. The machine learning-based adaptive viscous damping wall according to claim 5, wherein an outer cylindrical surface of the opening end of the outer sleeve (42) is fixedly connected with a cone-shaped dust cover (50) through a metal clamp, the other end of the dust cover (50) is fastened on the outer surface of the supporting shaft (45) through another metal clamp, the dust cover (50) is made of thermoplastic rubber, a closed flexible cavity is formed between the outer sleeve (42) and the supporting shaft (45), and lubricating grease is arranged in the flexible cavity.
7. The machine learning-based self-adaptive viscous damping wall is characterized in that the self-adaptive mechanism (4) further comprises a stress ring (51) fixedly sleeved on the outer surface of the middle of the outer sleeve (42), a limiting groove (52) communicated with the through hole (41) is formed in the viscous plate (32), a buffer groove (53) symmetrically distributed is formed in the viscous plate (32), the outer end of the buffer groove (53) is communicated with the through hole (41) and the limiting groove (52), a butterfly spring (54) is fixedly connected to the inner wall of one side of the buffer groove (53), and the outer surface of one side of the stress ring (51) is wrapped and fixedly connected with the free end of the butterfly spring (54).
8. The machine learning-based adaptive viscous damping wall is characterized in that the tackifying mechanism (6) comprises wing plates (61) fixedly connected to the outer surfaces of two free ends of the supporting shaft (45), the two wing plates (61) are symmetrically distributed around the center point of the supporting shaft (45) in a 180-degree rotation mode, transition round corner treatment is conducted on the edges of the outer surfaces of the wing plates (61), ball grooves (62) are formed in the upper surface and the lower surface of the wing plates (61) and are regularly distributed on the surfaces of the wing plates, cylindrical grooves (63) are fixedly communicated between the adjacent ball grooves (62), and therefore network-shaped communication grooves in a cross shape are formed in the upper surface and the lower surface of the wing plates (61).
9. The adaptive viscous damping wall based on machine learning of claim 8, wherein the viscous damping wall comprises a sensor system and an intelligent control system, the sensor system comprising: a displacement sensor (34) and an acceleration sensor (33) which are arranged on the damping body (3) and are used for monitoring the motion displacement, the speed and the acceleration of the damping body (3) in viscous liquid in real time; A temperature sensor (13) arranged on the inner wall of the channel steel box (1); A pressure sensor (35) mounted on the surface of the viscous plate (32); A strain sensor (21) mounted on the damper connecting wall pier (2); The intelligent control system comprises: the data acquisition module is connected with all the sensors and is used for acquiring sensor data in real time; The processor is internally provided with an optimization algorithm based on machine learning and is used for analyzing the state of the damping wall according to the sensor data and outputting an optimization instruction; And the control module is used for controlling the electrohydraulic servo valve of the lifting hydraulic cylinder (11) according to the instruction output by the processor so as to adjust the viscosity depth and is connected to the viscous liquid supply system so as to dynamically adjust the viscosity ratio of the viscous liquid.
10. The adaptive viscous damping wall based on machine learning of claim 9, wherein the optimization algorithm built in the processor is a hybrid control algorithm based on fusion of a long-term memory network and model predictive control, and the core calculation formula of the algorithm comprises: Step one, defining a system state vector: Wherein: as a system state vector of the system, In order to damp the displacement of the body, In order to damp the speed of the body, In order to dampen the acceleration of the body, For the temperature of the viscous liquid, For the average pressure on the surface of the viscous plate, Strain for the wall pier; Step two, defining a control instruction vector: Wherein: In order to control the vector of the instruction, In order to be a viscous depth adjustment command, A viscous liquid viscosity adjustment instruction; thirdly, an LSTM state prediction model: Wherein: in order to predict the state of the device, In order to predict the function of the object, In order to be in a hidden state, In the state of a cell, the cell is in a state of being, Is a network parameter; Fourth, model predictive control objective function: Wherein: The future control sequences to be optimized, The objective function is optimized and the method is characterized in that, The ideal reference state is adopted to be the ideal reference state, The error of the state is calculated and the error of the state is calculated, The weight matrix of the control quantity is a positive and negative angle matrix and is used for balancing tracking precision and control cost, The highest allowable working temperature of the viscous liquid, Representing at the current time Future predicted based on LSTM network model (and current known information The system state vector of the moment in time, The temperature overrun punishment coefficient is adopted, Representing at the current time Is the future th The control command vector planned for a control cycle, Representing the predicted temperature of the viscous liquid for the future first control period at the current moment; Step five, a viscosity dynamic adjustment formula: , wherein, For the viscosity compensation amount based on the constitutive model, In order to be based on the output of the rule base, Is the root mean square of the acceleration, In order to estimate the magnitude of the vibration, 、 Is a weighting coefficient.

Description

Self-adaptive viscous damping wall based on machine learning Technical Field The invention relates to the technical field of damping walls, in particular to a self-adaptive viscous damping wall based on machine learning. Background Damping walls, also known as damper walls or energy dissipating walls, are a high-efficiency passive control device used for resisting wind loads and earthquake actions in modern high-rise buildings, large-span bridges and important structures. Its core function is to consume energy of the input structure, reduce vibration and deformation of the structure, thereby improving comfort and safety of the structure. However, the prior damping wall still has the remarkable problem that under the action of medium or large vibration, the temperature of the viscous liquid is increased due to continuous shearing. If the temperature rises too fast or the heat dissipation is not good, viscosity of the viscous liquid can be greatly reduced, damping force is attenuated, energy consumption capability is obviously reduced, structural safety is affected, and meanwhile, damping plates of the traditional damping wall are mainly fixedly installed. Seismic excitation is multi-dimensional in nature, and besides the main vibration direction, strong transverse force can cause unexpected bending or twisting deformation of the damping plate, fatigue damage and even failure can be caused under long-term action, service life and reliability are affected, and the damping characteristics of the conventional device are fixed after the conventional device is installed in a passive working mode. The invention solves the defects of the prior art because the real-time intensity, the frequency spectrum characteristic and the actual response state of the structure of the earthquake are not dynamically adjusted, the optimal energy consumption state is difficult to be kept in the earthquake events with different earthquake magnitudes and different characteristics, and the hoisting space of the shock absorption efficiency is limited. Disclosure of Invention Based on the existing technical problems, the invention provides a self-adaptive viscous damping wall based on machine learning. The invention provides a self-adaptive viscous damping wall based on machine learning, which comprises a channel steel box internally loaded with viscous liquid, wherein the channel steel box is arranged between two damper connecting wall piers, a damping body is vertically arranged in the channel steel box, the upper surface of the damping body is fixedly connected with the lower surface of the upper damper connecting wall pier through an embedded part, a self-adaptive mechanism is arranged in the damping body in a penetrating way, two ends of the self-adaptive mechanism are provided with tackifying mechanisms, the channel steel box is made of high-strength steel and has good rigidity and corrosion resistance, the damping body is made of metal, and the surface of the damping body can be subjected to rust prevention treatment. The self-adaptive mechanism is used for keeping self-balance on the damping body in the moving process, wherein the damping body reciprocates along the length direction of the damper connecting wall pier through viscous liquid. The tackifying mechanism is fully contacted with the viscous liquid in the groove steel box, and is bonded and damped during vibration. Preferably, the lower surface of the channel steel box is symmetrically distributed and hinged with a lifting hydraulic cylinder, the piston rod end of the lifting hydraulic cylinder is hinged with the upper surface of the damper connecting wall pier through another embedded part, a plurality of buffer telescopic rods for lifting the channel steel box are fixedly arranged on the lower surface of the channel steel box through other groups of embedded parts, and the lower surface of the buffer telescopic rods is fixedly arranged on the upper surface of the damper connecting wall pier through corresponding embedded parts. Through above-mentioned technical scheme, in order to adapt to the target damping force, need adjust viscous degree of depth, then lift pneumatic cylinder and buffering telescopic link collaborative work, can adjust the cell steel box height in real time when the earthquake, adapt to different vibrations intensity, the buffering telescopic link provides elastic support simultaneously, avoids rigid impact, extension device life. Preferably, the damping body comprises horizontal connection roof and fixed connection be in connect the vertical viscous board of roof lower surface, connect the upper surface of roof through built-in fitting and upper portion the lower surface fixed connection of wall mound is connected to the attenuator, all there is the flow clearance between the week side surface of viscous board, lower surface and the inner wall of channel-section steel box. Through above-mentioned technical scheme, in order to realize the damping ef