DEVELOPMENT AND PRACTICE OF SIMULTANEOUS OPTIMUM DESIGN OF STRUCTURE AND CONTROL SYSTEM

本研究提出了新的结构和控制系统优化设计方法。提出了一种基于遗传算法的结构拓扑优化方法。通过优化结构的拓扑结构，使由固有频率组成的适应度函数最大化。通过优化有限元分析中有限元的位置来确定结构的最佳拓扑。在基于遗传算法的优化设计中，为了便于实际应用，应减少计算机的负担。本文对遗传算法中的交叉和变异操作进行了改进，以提高计算效率。此外，在结构分析中通过模型降阶，提高了计算效率。通过对磁盘装置中导向臂的拓扑优化，验证了该方法的有效性和实用性。其次，本研究提出结构与控制系统同时最佳化设计方法，利用模态分析与灵敏度分析，设计出由连续体结构所组成之伺服系统最佳化。首先用有限元法对机械结构进行建模，这种空间模型具有很高的自由度，无法设计控制系统。为了减少自由度，将空间模型转化为模态模型，并与动态补偿器构成伺服系统。然而，控制系统的稳定性并不总是得到保证，因为模态模型忽略了高阶模式。提出了一种结构和控制系统同时优化的方法，保证了实际系统的稳定性。采用灵敏度分析和非线性规划进行优化。仿真和实验验证了该方法的有效性。

This study proposes new optimum design methods of structure and control system. Forst, a structural topology optimization method based on the genetic argorithm is presented. The fitness function composed of a natural frequency is maximized by optimizing the topology of the structure. The optimum topology of the structure is determined by optimizing the placement of the finite elements in FEM analysis. In the optimum design based on the genetic argorithm, the computer burden should be decreased for practical use. This study improves the genetic operations with respect to the crossover and the mutation in order to improve the efficiency of the calculation. Furthermore, the higher efficiency of the calculation is achieved by the model reduction in structural analysis. Effectiveness and practicability of the proposed approach are verified by optimizing the topology of the guide-arm installed in the magnetic disk device. Next, this study proposes a sumultaneous optimum design method of structure and control system by which to design an optimum servosystem composed of continuous body structures using modal analysis and sensitivity analysis. The mechanical structures is modeled first by FEM.This spatial model are of such high degree of freedom that a control system cannot be designed. This spatial model is transformed into a modal model in order to reduce DOF.The servosystem is composed with the reduced model and a dynamic compensator. However, stability of the control system is not always guaranteed because the modal model ignores high-order modes. An simultaneous optimization method of the structure and the control system which guarantees stability of the actual system is proposed in this study. The optimization is performed using sensitivity analysis and nonlinear programming. The effectiveness of the proposed method is verified by both simulation and experiment.