Junichi Hongu

Structural vibration control by tuned mass damper using central pattern generator

This paper proposes a new control method for active mass dampers using a Central Pattern Generator in vibration mitigation. The active mass dampers (or active dynamic absorbers) have been applied to structural vibration control of high-rise buildings, bridges and so on. In this case, the mass of the active mass damper must oscillate in an appropriate phase in relation to the control object, and generally, the damper has been designed by linear control theory as pole placement method, optimal control method or H infinity control method, and all the rest. On the other hand, on walking of animate beings like mammals or insects, both side feet have appropriate phase relations; moreover, it is possible to keep moving on irregular ground. That is, algorithms for the walking would be embedded into the animate beings to control the complicated and redundant bodies with ease and robustness. In biological study, the Central Pattern Generators in bodies playing a significant role in the walking have been learned over the last few decades, and some studies said that some animate beings are able to control their feet by using the generators without their brains in the walking. Moreover, mathematical models of the pattern generators have been proposed, and some researchers have been studying to realize walking of biped-robots using the pattern generators embedded in a computer. In this study, the algorithm is installed into a controller for the active mass damper; furthermore, validation of the controller is performed by numerical simulation.

Mutual synchronization between structure and central pattern generator

This paper shows an evaluating method of synchronization between a structure and Central Pattern Generators (CPGs), which are embedded in a controller designed for an active mass damper. A neural oscillator composing the CPGs has nonlinear and entrainment properties. Therefore, the proposed controller has possibility to exhibit the characteristic of robustness, when the structural parameters, i.e. stiffness or damping, are changed by earthquakes and the like. Our earlier studies have proposed the new controller and ascertained the efficacy of vibration suppression. However, there has been no study to evaluate the controllers above-mentioned properties. For tuning into practical application, the reliability and robustness along with the controllers vibration mitigation performance must be analyzed. In this paper, phase reduction theory is tried to appraise the synchronization between a structure and the CPGs. In this case, the synchronization between the target structure and a single neural oscillator constituting the CPGs is required to be investigated. Therefore, the single neural oscillators the harmonization characteristic with sinusoidal input is firstly examined, and the synchronization region is expressed using phase response curves. In addition, the mutual synchronization between the structure and the single neural oscillator is studied under sinusoidal input using the result of the harmonization characteristic.

Evaluation method for a controller of active mass damper using central pattern generator

This paper proposes an evaluation method for a CPG controller designed for active mass dampers. Neural oscillators composing the CPG have nonlinear and entrainment properties. Therefore, the proposed controller has possibility to have flexibility, when the structural parameters, i.e. stiffness or damping, are changed by the effect of earthquakes and the like. However, there has been no study to evaluate the controller’s above-mentioned properties. For tuning into practical application, the reliability and flexibility along with the controller’s performance must be analyzed. In our previous study, the phase reduction theory was tried to appraise the synchronization between a structure and a single neural oscillator and the synchronization region of the neural oscillator was obtained as basic research. However, the information from the synchronization region was insufficient to evaluate the system, because the neural oscillator has a phase difference called a phase locking point between the structure and the neural oscillator during the synchronization. Then, in this paper, the phase locking point within the synchronization region between a structure and a single neural oscillator is focused on, and the phase locking point and the vibration mitigation effect are considered with the simple object model.

Output analysis of swarm of neural oscillators stimulated by earthquake-induced acceleration responses of a structure

This paper shows input-output analysis of a neural oscillator swarm stimulated by earthquake-induced acceleration responses of a structure. We have proposed a new active mass damper system consisting of a neural oscillator and a position controller. However, the proposed system has not adapted successfully to parameter changes of the structure. Recent studies in biology have demonstrated that multiple oscillators have hierarchical network structures to ensure adaptation to environmental changes. To improve the robust performance of the proposed system by constructing of hierarchical network of neural oscillators, there is a need for a better understanding of nature of different neural oscillators. This research addressed this need by visualizing output of swarm of neural oscillators, whose natural frequencies and input gains are different. The numerical information of output is visualized by grayscale, and the relation of output of different neural oscillators is considered when input is the same. As a result, the research provides new information that predicts the instant center frequency of a structure excited by earthquakes.

Active mass damper system for high-rise buildings using neural oscillator and position controller: generation method for desired stroke of auxiliary mass using synchronous detection

Reducing vibration of high-rise structures under earthquake load has been the subject of considerable efforts in Japan. Relevant researches about vibration energy dissipation devices for buildings have been undertaken. An active mass damper is one of the well-known vibration control devices. Despite the accumulation of much knowledge of control design methods for the system, application of the devices to high-rise structures under earthquake load is challenging, because the active mass dampers have one serious disadvantage about stroke limitation of the auxiliary mass. In this study, we have proposed a new control system, which had a neural oscillator and position controller, to solve this problem. The main role of this neural oscillator included in newly proposed system is picking up the phase information of the eigen-frequency component of a target structure, then the auxiliary mass of an active mass damper is excited by reference to the oscillator’s signal. We can easily regulate the stroke of the active mass damper no matter how large the target structure swings, because the control signal for the auxiliary mass of the phase and amplitude information of the active mass damper are separately processed. However, there is no general determination method for the desired stroke of the auxiliary mass from the oscillator’s signal. The previous method determined the desired stroke of the auxiliary mass using two state quantities of the oscillator, which depends on types of oscillators and has non-linearity and instability. Thus, this study proposes a generation method for the desired stroke of the auxiliary mass by using synchronous detection. From the results of numerical simulation, the presenting method can apply to any types of oscillators and generate the linear and stable signal by reference to an oscillators’ signal, and was effective for improving the control performance.

Active mass damper system for high-rise buildings using neural oscillator and position controller: sinusoidally varying desired displacement of auxiliary mass intended for reduction of maximum control force

Reducing vibration of high-rise structures under earthquake load has been the subject of considerable efforts in Japan. Relevant researches about vibration energy dissipation devices for buildings have been undertaken. An active mass damper is one of the well-known vibration control devices. Despite the accumulation of much knowledge of control design methods for the system, application of the devices to high-rise structures under earthquake load is challenging, because the active mass dampers have one serious disadvantage about stroke limitation of the auxiliary mass. In this study, we have proposed a new control system, which had a neural oscillator and position controller, to solve this problem. The objective of this paper is to improve the vibration control performance of the proposed active mass damper system. The previous method generated rectangular waves as the desired displacement, whose amplitude is varied in accordance with the vibration responses of a structure excited by earthquakes. Furthermore, the gains of the position controller, which derives the auxiliary mass to the desired displacement, have been designed in consideration of response reduction of the structure. However, the generated rectangular desired displacement was not adequate to reduce the maximum acceleration responses of the structure, because the driving force for the auxiliary mass generates excessive amounts of acceleration as the direction of the desired displacement is switched. Thus, this paper proposes a new method, which generates sinusoidal varying desired displacement for the auxiliary mass of the active mass damper system to reduce the acceleration response of structures. The results of numerical simulation showed that the proposed method in this work was effective for improving the control performance.

Active mass damper system for high-rise buildings using neural oscillator and position controller considering stroke limitation of the auxiliary mass

This paper proposes a problem-solving method for the stroke limitation problem, which is related to auxiliary masses of active mass damper systems for high-rise buildings. The proposed method is used in a new simple control system for the active mass dampers mimicking the motion of bipedal mammals, which has a neural oscillator synchronizing with the acceleration response of structures and a position controller. In the system, the travel distance and direction of the auxiliary mass of the active mass damper is determined by reference to the output of the neural oscillator, and then, the auxiliary mass is transferred to the decided location by using a PID controller. The one of the purpose of the previouslyproposed system is stroke restriction problem avoidance of the auxiliary mass during large earthquakes by the determination of the desired value within the stroke limitation of the auxiliary mass. However, only applying the limited desired value could not rigorously restrict the auxiliary mass within the limitation, because the excessive inertia force except for the control force produced by the position controller affected on the motion of the auxiliary mass. In order to eliminate the effect on the auxiliary mass by the structural absolute acceleration, a cancellation method is introduced by adding a term to the control force of the position controller. We first develop the previously-proposed system for the active mass damper and the additional term for cancellation, and verity through numerical experiments that the new system is able to operate the auxiliary mass within the restriction during large earthquakes. Based on the comparison of the proposed system with the LQ system, a conclusion was drawn regarding which the proposed neuronal system with the additional term appears to be able to limit the stroke of the auxiliary mass of the AMD.

Experimental evaluation of a control system for active mass dampers consisting of a position controller and neural oscillator

This paper shows experimental performance evaluation of a new control system for active mass dampers (AMDs). The proposed control system consists of a position controller and neural oscillator, and is designed for the solution of a stroke limitation problem of an auxiliary mass of the AMDs. The neural oscillator synchronizing with the response of a structure generates a signal, which is utilized for switching of motion direction of the auxiliary mass and for travel distances of the auxiliary mass. According to the generated signal, the position controller drives the auxiliary mass to the target values, and the reaction force resulting from the movement of the auxiliary mass is transmitted to the structure, and reduces the vibration amplitude of the structure. Our previous research results showed that the proposed system could reduce the vibration of the structure while the motion of auxiliary mass was suppressed within the restriction; however the control performance was evaluated numerically. In order to put the proposed system to practical use, the system should be evaluated experimentally. This paper starts by illustrating the relation among subsystems of the proposed system, and then, shows experimental responses of a structure model with the AMD excited by earthquakes on a shaker to confirm the validity of the system.

A driven active mass damper by using output of a neural oscillator (effects of position control system changes on vibration mitigation performance)

In this paper, a design method for a PD controller, which is a part of a new active mass damper system using a neural oscillator for high-rise buildings, is proposed. The new system mimicking the motion of bipedal mammals is a quite simple system, which has the neural oscillator synchronizing with the acceleration response of the structure. The travel distance and direction of the auxiliary mass of the active mass damper is decided by the output of the neural oscillator, and then, the auxiliary mass is transferred to the decided location by using the PD controller. Therefore, the performance of the PD controller must be evaluated by the vibration energy absorbing efficiency by the system. In order to bring the actual path driven by the PD controller in closer alignment with the ideal path, which is assumed to be a sinusoidal wave under resonance, firstly, the path of the auxiliary mass driven by the PD controller is analytically derived, and the inner product between the vector of ideal and analytical path is evaluated. And then, the PD gain is decided by the maximum value of the inner product. Finally, numerical simulations confirm the validity of the proposed design method of the PD controller.

Experimental evaluation of a neural-oscillator-driven active mass damper system

This paper proposes a new active dynamic absorber control system for high-rise buildings using a neural oscillator and a map, which estimates the amplitude level of the oscillator, and shows some experimental results by using an apparatus, which realizes the proposed control algorithm. The proposed system decides the travel distance and direction of the auxiliary mass of the dynamic absorber using the output of oscillator, which is the filtering result of structure acceleration responses by the property of the oscillator, and Amplitude-Phase map (AP-map) for estimation of the structural response in specific frequency between synchronization region, and then, transfer the auxiliary mass to the predetermined location by using a position controller. In addition, the developed active dynamic absorber system is mounted on the top of the experimental single degree of freedom structure, which represents high-rise buildings, and consists of the auxiliary mass, a DC motor, a ball screw, a microcomputer, a laser displacement sensor, and an acceleration sensor. The proposed AP-map and the algorithm to determine the travel direction of the mass using the oscillator output are embedded in the microcomputer. This paper starts by illuminating the relation among subsystems of the proposed system with reference to a block diagram, and then, shows experimental responses of the whole system excited by earthquakes to confirm the validity of the proposed system.