基于神经网络—遗传算法的双层隔振系统复合振动控制

Composite Vibration Control for Two-stage Vibration Isolating System Using Neural Network-Genetic Algorithm

针对实际应用中双层隔振系统参数时变,不易建模的特性,提出了将遗传算法及神经网络应用于双层隔振系统的复合振动控制策略。该方法利用遗传算法在线计算作用于隔振系统上的控制力,利用神经网络模拟隔振系统的动力特性,代替隔振系统进行动力分析,该系统充分发挥了遗传算法及神经网络各自的优点。其中,遗传算法采用与单纯形相结合的混合遗传算法,结合了遗传算法良好的全局收敛性和单纯形算法的优秀的局部搜索能力,提高了搜索速度与精度,神经网络采用了一种学习速率可自适应调整的BP算法,提高了神经网络的收敛速度。仿真结果表明了该方法的有效性。

In the paper the vibrating characters of two-stage vibration isolating system is analyzed.A new active vibration control system,the genetic algorithm-neural network system is proposed.In this system the genetic algorithm is used to determine the control force and mixture genetic algorithm fusing the advantage of steeply convergence and highly precession of simplex algorithm and global convergence of genetic algorithm.The algorithm improves the searching process and precision.The BP neural network＇s learning rate can be adjusted adaptively,is applied to simulate vibrating characters of vibration isolating system.The system well combines the genetic algorithm controller and neural network identifier.Simulation results show the effectiveness of the proposed method.