基于PTCNN的结构布局优化问题研究

Pulse transiently chaotic neural network for layout optimization

针对结构优化中的难点布局优化问题,基于交叉学科的优势,运用脉冲暂态混沌神经网络(PTCNN)的方法,对一个新型布局优化模型进行了寻优计算。在PTCNN算法中应用脉冲混沌动力自然解决了杆件的恢复和删除问题;利用神经网络的多神经元并行构成特点,通过参数调整方法解决了不同数量级变量间的耦合问题。因此PTCNN不仅是模型的寻优算法,更成为布局优化的一部分。此外构造了一个应力关联系数σAI,使面积变量随杆中应力大小按比例自适应下降。算例结果表明,基于PTCNN方法解决结构布局优化问题,有效且具有自适应性,布局优化效果明显。

Pulse Transiently Chaotic Neural Network （PTCNN） is used to solve layout optimization.The pole can automatically resume because of the abundance character of the arithmetic. The problem of coupling between different variables is solved by neural network＇s characters that the neural network is made up of many nerve cell. During the process of finding the optima, stress relating coefficient is adopted, which can make the area variables descent adaptably according to stress level. It is showed by an example that the method in this paper is efficient.