Yoyong Arfiadi

Optimal direct (static) output feedback controller using real coded genetic algorithms

This paper considers a genetic based optimisation procedure for designing the control force in actively controlled structures. Two optimal output feedback controllers utilising H2 and H∞ optimal control are discussed in this paper where only a limited number of sensors is used. In addition, the feedback control used in this paper is a static (direct) output feedback control where the gain can be multiplied directly with the measurement output. Therefore, it is not necessary to construct an observer to estimate the state as in the case of dynamic or observer based controller. The optimisation to minimise the H2 and H∞ norms are solved using real coded genetic algorithms, which have the ability to explore the unknown domain of the solution. Numerical examples are then carried out to the structures utilising active tuned mass damper and active bracing systems to show the applicability of the proposed procedure.

Passive and active control of three‐dimensional buildings

The majority of the recent research effort on structural control considers two-dimensional plane structures. However, not all buildings can be modelled as plane structures, thus limiting the capability of the proposed procedures only to regular and symmetrical structures. A new procedure is developed in this paper to analyse three-dimensional buildings utilizing passive and active control devices. In the building model, the floors are assumed rigid in their own plane resulting in three degrees of freedom at each floor. Two types of active control devices utilizing an active tuned mass damper and an active bracing system are considered. The effect of passive mass dampers and active control force in the equations of motion is incorporated by using the Hamiltons principle. The passive parameters of the dampers as well as the controller gain is then optimized using a genetic based optimizer where the H 2 , H∞ and L 1 norms are taken as the objective functions.

Optimum rigid pavement design by genetic algorithms

Abstract The design of rigid pavements according to AUSTROADS involves assuming a pavement structure then using a number of tables and figures to calculate the two governing design criteria, the flexural fatigue of the concrete base and the erosion of the sub-grade/sub-base. Each of these two criteria needs to be less than 100%. The designer needs to ensure that both criteria are near 100% so that safe and economical designs are achieved. This paper presents a formulation for the problem of optimum rigid road pavement design by defining the objective function, which is the total cost of pavement materials, and all the constraints that influence the design. A genetic algorithm is used to find the optimum design. The results obtained from the genetic algorithm are compared with results obtained from a Newton–Raphson based optimisation solver. The latter being developed using spreadsheets.

Using genetic algorithm for optimal control force in active vibration control

In this paper the procedure of genetic algorithm (GA) is used to solve the optimal control force in active structural control. Two types of performance indices are examined in this paper. The first performance index is a quadratic regulator and the second is the H? norm of the transfer function. The direct (static) output feedback is used for both cases, instead of full state or dynamic output feedback. Numerical simulations are then carried out to a single and multi-degree-of-freedom structure with active mass damper to show the effectiveness of the GA procedure. It is shown that by using GA procedure the determination of optimal control force with direct output feedback is highly simplified.